

- 1.Reinders, C., Schubert, F., and Rosenhahn, B. (2022) ChimeraMix: Image Classification on Small Datasets via Masked Feature Mixing. In Arxiv Preprint.
@inproceedings{ReiSch2022,
author = {Reinders, Christoph and Schubert, Frederik and Rosenhahn, Bodo},
booktitle = {Arxiv Preprint},
keywords = {Classification Feature Image Masked Mixing leibnizailab myown},
month = {mar},
title = {ChimeraMix: Image Classification on Small Datasets via Masked Feature Mixing},
year = 2022
}%0 Conference Paper
%1 ReiSch2022
%A Reinders, Christoph
%A Schubert, Frederik
%A Rosenhahn, Bodo
%B Arxiv Preprint
%D 2022
%T ChimeraMix: Image Classification on Small Datasets via Masked Feature Mixing - 1.Awiszus, M., Schubert, F., and Rosenhahn, B. (2022) Wor(l)d-GAN: Towards Natural Language Based PCG in Minecraft, IEEE Transactions on Games.
@article{AwiSch2022,
author = {Awiszus, Maren and Schubert, Frederik and Rosenhahn, Bodo},
journal = {IEEE Transactions on Games},
keywords = {Based Language Minecraft PCG in leibnizailab myown},
month = {feb},
note = {11 pages, 10 figures.},
title = {Wor(l)d-GAN: Towards Natural Language Based PCG in Minecraft},
year = 2022
}%0 Journal Article
%1 AwiSch2022
%A Awiszus, Maren
%A Schubert, Frederik
%A Rosenhahn, Bodo
%D 2022
%J IEEE Transactions on Games
%T Wor(l)d-GAN: Towards Natural Language Based PCG in Minecraft - 1.Grimm, E., Kuhnke, F., Gajdt, A., Ostermann, J., and Knoche, M. (2022) Accurate Quantification of Anthocyanin in Red Flesh Apples Using Digital Photography and Image Analysis, Horticulturae 8.
@article{GriKuh2022,
author = {Grimm, Eckhard and Kuhnke, Felix and Gajdt, Anna and Ostermann, Jörn and Knoche, Moritz},
journal = {Horticulturae},
keywords = {Anthocyanin Apples Flesh Quantification Red in leibnizailab myown of},
month = {jan},
number = 2,
title = {Accurate Quantification of Anthocyanin in Red Flesh Apples Using Digital Photography and Image Analysis},
volume = 8,
year = 2022
}%0 Journal Article
%1 GriKuh2022
%A Grimm, Eckhard
%A Kuhnke, Felix
%A Gajdt, Anna
%A Ostermann, Jörn
%A Knoche, Moritz
%D 2022
%J Horticulturae
%N 2
%R https://doi.org/10.3390/horticulturae8020145
%T Accurate Quantification of Anthocyanin in Red Flesh Apples Using Digital Photography and Image Analysis
%U https://www.mdpi.com/2311-7524/8/2/145
%V 8 - 1.Hvarfner, C., Stoll, D., Souza, A., Nardi, L., Lindauer, M., and Hutter, F. (2022) piBO: Augmenting Acquisition Functions with User Beliefs for Bayesian Optimization. In 10th International Conference on Learning Representations, ICLR’22, pp. 1–30.
@inproceedings{HvaSto2022a,
author = {Hvarfner, Carl and Stoll, Danny and Souza, Artur and Nardi, Luigi and Lindauer, Marius and Hutter, Frank},
booktitle = {10th International Conference on Learning Representations, ICLR'22},
keywords = {Acquisition Augmenting Functions leibnizailab myown},
month = {apr},
pages = {1-30},
title = {piBO: Augmenting Acquisition Functions with User Beliefs for Bayesian Optimization},
year = 2022
}%0 Conference Paper
%1 HvaSto2022a
%A Hvarfner, Carl
%A Stoll, Danny
%A Souza, Artur
%A Nardi, Luigi
%A Lindauer, Marius
%A Hutter, Frank
%B 10th International Conference on Learning Representations, ICLR'22
%D 2022
%P 1-30
%T piBO: Augmenting Acquisition Functions with User Beliefs for Bayesian Optimization
%U https://openreview.net/pdf/1ce81b811a1cac6ed2405793a93e8512b1b50005.pdf - 1.Hinrichs, R., Gerkens, K., Lange, A., and Ostermann, J. (2022) Classification of Guitar Effects and Extraction of their Parameter Settings from Instrument Mixes Using Convolutional Neural Networks. In EvoMUSART 2022.
@inproceedings{HinGer2022,
author = {Hinrichs, Reemt and Gerkens, Kevin and Lange, Alexander and Ostermann, Jörn},
booktitle = {EvoMUSART 2022},
keywords = {Classification Effects Extraction Guitar and leibnizailab myown of},
title = {Classification of Guitar Effects and Extraction of their Parameter Settings from Instrument Mixes Using Convolutional Neural Networks},
year = 2022
}%0 Conference Paper
%1 HinGer2022
%A Hinrichs, Reemt
%A Gerkens, Kevin
%A Lange, Alexander
%A Ostermann, Jörn
%B EvoMUSART 2022
%D 2022
%T Classification of Guitar Effects and Extraction of their Parameter Settings from Instrument Mixes Using Convolutional Neural Networks - 1.Mukherjee, R., Vishnu, U., Peruri, H. C., Bhattacharya, S., Rudra, K., Goyal, P., and Ganguly, N. (2022) MTLTS: A Multi-Task Framework To Obtain Trustworthy Summaries From Crisis-Related Microblogs. In Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining, pp. 755–763, Association for Computing Machinery, Virtual Event, AZ, USA.Occurrences of catastrophes such as natural or man-made disasters trigger the spread of rumours over social media at a rapid pace. Presenting a trustworthy and summarized account of the unfolding event in near real-time to the consumers of such potentially unreliable information thus becomes an important task. In this work, we propose MTLTS, the first end-to-end solution for the task that jointly determines the credibility and summary-worthiness of tweets. Our credibility verifier is designed to recursively learn the structural properties of a Twitter conversation cascade, along with the stances of replies towards the source tweet. We then take a hierarchical multi-task learning approach, where the verifier is trained at a lower layer, and the summarizer is trained at a deeper layer where it utilizes the verifier predictions to determine the salience of a tweet. Different from existing disaster-specific summarizers, we model tweet summarization as a supervised task. Such an approach can automatically learn summary-worthy features, and can therefore generalize well across domains. When trained on the PHEME dataset [29], not only do we outperform the strongest baselines for the auxiliary task of verification/rumour detection, we also achieve 21 - 35% gains in the verified ratio of summary tweets, and 16 - 20% gains in ROUGE1-F1 scores over the existing state-of-the-art solutions for the primary task of trustworthy summarization.
@inproceedings{10.1145/3488560.3498536,
abstract = {Occurrences of catastrophes such as natural or man-made disasters trigger the spread of rumours over social media at a rapid pace. Presenting a trustworthy and summarized account of the unfolding event in near real-time to the consumers of such potentially unreliable information thus becomes an important task. In this work, we propose MTLTS, the first end-to-end solution for the task that jointly determines the credibility and summary-worthiness of tweets. Our credibility verifier is designed to recursively learn the structural properties of a Twitter conversation cascade, along with the stances of replies towards the source tweet. We then take a hierarchical multi-task learning approach, where the verifier is trained at a lower layer, and the summarizer is trained at a deeper layer where it utilizes the verifier predictions to determine the salience of a tweet. Different from existing disaster-specific summarizers, we model tweet summarization as a supervised task. Such an approach can automatically learn summary-worthy features, and can therefore generalize well across domains. When trained on the PHEME dataset [29], not only do we outperform the strongest baselines for the auxiliary task of verification/rumour detection, we also achieve 21 - 35% gains in the verified ratio of summary tweets, and 16 - 20% gains in ROUGE1-F1 scores over the existing state-of-the-art solutions for the primary task of trustworthy summarization.},
address = {New York, NY, USA},
author = {Mukherjee, Rajdeep and Vishnu, Uppada and Peruri, Hari Chandana and Bhattacharya, Sourangshu and Rudra, Koustav and Goyal, Pawan and Ganguly, Niloy},
booktitle = {Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining},
keywords = {leibnizailab myown},
pages = {755–763},
publisher = {Association for Computing Machinery},
series = {WSDM '22},
title = {MTLTS: A Multi-Task Framework To Obtain Trustworthy Summaries From Crisis-Related Microblogs},
year = 2022
}%0 Conference Paper
%1 10.1145/3488560.3498536
%A Mukherjee, Rajdeep
%A Vishnu, Uppada
%A Peruri, Hari Chandana
%A Bhattacharya, Sourangshu
%A Rudra, Koustav
%A Goyal, Pawan
%A Ganguly, Niloy
%B Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining
%C New York, NY, USA
%D 2022
%I Association for Computing Machinery
%P 755–763
%R 10.1145/3488560.3498536
%T MTLTS: A Multi-Task Framework To Obtain Trustworthy Summaries From Crisis-Related Microblogs
%U https://doi.org/10.1145/3488560.3498536
%X Occurrences of catastrophes such as natural or man-made disasters trigger the spread of rumours over social media at a rapid pace. Presenting a trustworthy and summarized account of the unfolding event in near real-time to the consumers of such potentially unreliable information thus becomes an important task. In this work, we propose MTLTS, the first end-to-end solution for the task that jointly determines the credibility and summary-worthiness of tweets. Our credibility verifier is designed to recursively learn the structural properties of a Twitter conversation cascade, along with the stances of replies towards the source tweet. We then take a hierarchical multi-task learning approach, where the verifier is trained at a lower layer, and the summarizer is trained at a deeper layer where it utilizes the verifier predictions to determine the salience of a tweet. Different from existing disaster-specific summarizers, we model tweet summarization as a supervised task. Such an approach can automatically learn summary-worthy features, and can therefore generalize well across domains. When trained on the PHEME dataset [29], not only do we outperform the strongest baselines for the auxiliary task of verification/rumour detection, we also achieve 21 - 35% gains in the verified ratio of summary tweets, and 16 - 20% gains in ROUGE1-F1 scores over the existing state-of-the-art solutions for the primary task of trustworthy summarization.
%@ 9781450391320 - 1.Chang, Y., Jing, X., Ren, Z., and Schuller, B. W. (2022) CovNet: A transfer learning framework for automatic COVID-19 detection from crowd-sourced cough sounds, Frontiers in Digital Health (Hochheiser, H., Ed.) 3, 1–11.
@article{chang2022covnet,
author = {Chang, Yi and Jing, Xin and Ren, Zhao and Schuller, Björn W.},
editor = {Hochheiser, Harry},
journal = {Frontiers in Digital Health},
keywords = {leibnizailab},
month = {Jan.},
number = 799067,
pages = {1--11},
title = {CovNet: A transfer learning framework for automatic COVID-19 detection from crowd-sourced cough sounds},
volume = 3,
year = 2022
}%0 Journal Article
%1 chang2022covnet
%A Chang, Yi
%A Jing, Xin
%A Ren, Zhao
%A Schuller, Björn W.
%D 2022
%E Hochheiser, Harry
%J Frontiers in Digital Health
%N 799067
%P 1--11
%R https://doi.org/10.3389/fdgth.2021.799067
%T CovNet: A transfer learning framework for automatic COVID-19 detection from crowd-sourced cough sounds
%V 3 - 1.Geisler, S., Vidal, M.-E., Cappiello, C., Loscio, B. F., Gal, A., Jarke, M., Lenzerini, M., Missier, P., Otto, B., Paja, E., Pernici, B., and Rehof, J. (2022) Knowledge-Driven Data Ecosystems Toward Data Transparency, Journal of Data and Information Quality, Association for Computing Machinery (ACM) 14, 1–12.
@article{Geisler_2022,
author = {Geisler, Sandra and Vidal, Maria-Esther and Cappiello, Cinzia and Loscio, Bernadette Farias and Gal, Avigdor and Jarke, Matthias and Lenzerini, Maurizio and Missier, Paolo and Otto, Boris and Paja, Elda and Pernici, Barbara and Rehof, Jakob},
journal = {Journal of Data and Information Quality},
keywords = {entitylinking knowledge-graphs leibnizailab myown semanticdataintegration},
month = {mar},
number = 1,
pages = {1--12},
publisher = {Association for Computing Machinery (ACM)},
title = {Knowledge-Driven Data Ecosystems Toward Data Transparency},
volume = 14,
year = 2022
}%0 Journal Article
%1 Geisler_2022
%A Geisler, Sandra
%A Vidal, Maria-Esther
%A Cappiello, Cinzia
%A Loscio, Bernadette Farias
%A Gal, Avigdor
%A Jarke, Matthias
%A Lenzerini, Maurizio
%A Missier, Paolo
%A Otto, Boris
%A Paja, Elda
%A Pernici, Barbara
%A Rehof, Jakob
%D 2022
%I Association for Computing Machinery (ACM)
%J Journal of Data and Information Quality
%N 1
%P 1--12
%R 10.1145/3467022
%T Knowledge-Driven Data Ecosystems Toward Data Transparency
%U https://doi.org/10.1145%2F3467022
%V 14 - 1.Benjamins, C., Eimer, T., Schubert, F., Mohan, A., Biedenkapp, A., Rosenhahn, B., Hutter, F., and Lindauer, M. (2022) Contextualize Me - The Case for Context in Reinforcement Learning, ArXiv Preprint.
@article{BenEim2022a,
author = {Benjamins, Carolin and Eimer, Theresa and Schubert, Frederik and Mohan, Aditya and Biedenkapp, André and Rosenhahn, Bodo and Hutter, Frank and Lindauer, Marius},
journal = {ArXiv Preprint},
keywords = {Case Context Learning Reinforcement for in leibnizailab myown},
month = {feb},
title = {Contextualize Me - The Case for Context in Reinforcement Learning},
year = 2022
}%0 Journal Article
%1 BenEim2022a
%A Benjamins, Carolin
%A Eimer, Theresa
%A Schubert, Frederik
%A Mohan, Aditya
%A Biedenkapp, André
%A Rosenhahn, Bodo
%A Hutter, Frank
%A Lindauer, Marius
%D 2022
%J ArXiv Preprint
%T Contextualize Me - The Case for Context in Reinforcement Learning
%U https://arxiv.org/abs/2202.04500 - 1.Adhisantoso, Y. G., and Ostermann, J. (2021) Method for the Coding of Contact Matrix m56622, ISO/IEC JTC 1/SC 29/WG 8.
@article{AdhO2021,
author = {Adhisantoso, Yeremia Gunawan and Ostermann, Jörn},
journal = {ISO/IEC JTC 1/SC 29/WG 8},
keywords = {Contact Matrix Method leibnizailab myown},
month = {apr},
title = {Method for the Coding of Contact Matrix m56622},
year = 2021
}%0 Journal Article
%1 AdhO2021
%A Adhisantoso, Yeremia Gunawan
%A Ostermann, Jörn
%D 2021
%J ISO/IEC JTC 1/SC 29/WG 8
%T Method for the Coding of Contact Matrix m56622 - 1.Narisetti, N., Henke, M., Seiler, C., Junker, A., Ostermann, J., Altmann, T., and Gladilin, E. (2021) Fully-automated root image analysis (faRIA), Scientific Reports 11.
@article{NarHen2021,
author = {Narisetti, Narendra and Henke, Michael and Seiler, Christiane and Junker, Astrid and Ostermann, Jörn and Altmann, Thomas and Gladilin, Evgeny},
journal = {Scientific Reports},
keywords = {analysis image leibnizailab myown root},
month = {aug},
title = {Fully-automated root image analysis (faRIA)},
volume = 11,
year = 2021
}%0 Journal Article
%1 NarHen2021
%A Narisetti, Narendra
%A Henke, Michael
%A Seiler, Christiane
%A Junker, Astrid
%A Ostermann, Jörn
%A Altmann, Thomas
%A Gladilin, Evgeny
%D 2021
%J Scientific Reports
%R https://doi.org/10.1038/s41598-021-95480-y
%T Fully-automated root image analysis (faRIA)
%U https://www.nature.com/articles/s41598-021-95480-y
%V 11 - 1.Voges, J., Hernaez, M., Mattavelli, M., and Ostermann, J. (2021) An Introduction to MPEG-G: The First Open ISO/IEC Standard for the Compression and Exchange of Genomic Sequencing Data, Proceedings of the IEEE 109, 1607–1622.
@article{VogHer2021a,
author = {Voges, Jan and Hernaez, Mikel and Mattavelli, Marco and Ostermann, Jörn},
journal = {Proceedings of the IEEE},
keywords = {Introduction MPEG-G leibnizailab myown to},
number = 9,
pages = {1607-1622},
title = {An Introduction to MPEG-G: The First Open ISO/IEC Standard for the Compression and Exchange of Genomic Sequencing Data},
volume = 109,
year = 2021
}%0 Journal Article
%1 VogHer2021a
%A Voges, Jan
%A Hernaez, Mikel
%A Mattavelli, Marco
%A Ostermann, Jörn
%D 2021
%J Proceedings of the IEEE
%N 9
%P 1607-1622
%R 10.1109/JPROC.2021.3082027
%T An Introduction to MPEG-G: The First Open ISO/IEC Standard for the Compression and Exchange of Genomic Sequencing Data
%U https://doi.org/10.1109/JPROC.2021.3082027
%V 109 - 1.Kaushal, A., Saha, A., and Ganguly, N. (2021) tWT–WT: A Dataset to Assert the Role of Target Entities for Detecting Stance of Tweets, pp. 3879–3889.The stance detection task aims at detecting the stance of a tweet or a text for a target. These targets can be named entities or free-form sentences (claims). Though the task involves reasoning of the tweet with respect to a target, we find that it is possible to achieve high accuracy on several publicly available Twitter stance detection datasets without looking at the target sentence. Specifically, a simple tweet classification model achieved human-level performance on the WT–WT dataset and more than two-third accuracy on various other datasets. We investigate the existence of biases in such datasets to find the potential spurious correlations of sentiment-stance relations and lexical choice associated with the stance category. Furthermore, we propose a new large dataset free of such biases and demonstrate its aptness on the existing stance detection systems. Our empirical findings show much scope for research on the stance detection task and proposes several considerations for creating future stance detection datasets.
@proceedings{J,
abstract = {The stance detection task aims at detecting the stance of a tweet or a text for a target. These targets can be named entities or free-form sentences (claims). Though the task involves reasoning of the tweet with respect to a target, we find that it is possible to achieve high accuracy on several publicly available Twitter stance detection datasets without looking at the target sentence. Specifically, a simple tweet classification model achieved human-level performance on the WT–WT dataset and more than two-third accuracy on various other datasets. We investigate the existence of biases in such datasets to find the potential spurious correlations of sentiment-stance relations and lexical choice associated with the stance category. Furthermore, we propose a new large dataset free of such biases and demonstrate its aptness on the existing stance detection systems. Our empirical findings show much scope for research on the stance detection task and proposes several considerations for creating future stance detection datasets.},
author = {Kaushal, Ayush and Saha, Avirup and Ganguly, Niloy},
keywords = {leibnizailab myown},
month = {jun},
pages = {3879-3889},
title = {tWT–WT: A Dataset to Assert the Role of Target Entities for Detecting Stance of Tweets},
year = 2021
}%0 Conference Proceedings
%1 J
%A Kaushal, Ayush
%A Saha, Avirup
%A Ganguly, Niloy
%D 2021
%P 3879-3889
%T tWT–WT: A Dataset to Assert the Role of Target Entities for Detecting Stance of Tweets
%X The stance detection task aims at detecting the stance of a tweet or a text for a target. These targets can be named entities or free-form sentences (claims). Though the task involves reasoning of the tweet with respect to a target, we find that it is possible to achieve high accuracy on several publicly available Twitter stance detection datasets without looking at the target sentence. Specifically, a simple tweet classification model achieved human-level performance on the WT–WT dataset and more than two-third accuracy on various other datasets. We investigate the existence of biases in such datasets to find the potential spurious correlations of sentiment-stance relations and lexical choice associated with the stance category. Furthermore, we propose a new large dataset free of such biases and demonstrate its aptness on the existing stance detection systems. Our empirical findings show much scope for research on the stance detection task and proposes several considerations for creating future stance detection datasets. - 1.Nandy, A., Sharma, S., Maddhashiya, S., Sachdeva, K., Goyal, P., and Ganguly, N. (2021) Question Answering over Electronic Devices: A New Benchmark Dataset and a Multi-Task Learning based QA Framework. In Findings of the Association for Computational Linguistics: EMNLP 2021, Association for Computational Linguistics.
@inproceedings{Nandy_2021,
author = {Nandy, Abhilash and Sharma, Soumya and Maddhashiya, Shubham and Sachdeva, Kapil and Goyal, Pawan and Ganguly, NIloy},
booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2021},
keywords = {leibnizailab myown},
publisher = {Association for Computational Linguistics},
title = {Question Answering over Electronic Devices: A New Benchmark Dataset and a Multi-Task Learning based QA Framework},
year = 2021
}%0 Conference Paper
%1 Nandy_2021
%A Nandy, Abhilash
%A Sharma, Soumya
%A Maddhashiya, Shubham
%A Sachdeva, Kapil
%A Goyal, Pawan
%A Ganguly, NIloy
%B Findings of the Association for Computational Linguistics: EMNLP 2021
%D 2021
%I Association for Computational Linguistics
%R 10.18653/v1/2021.findings-emnlp.392
%T Question Answering over Electronic Devices: A New Benchmark Dataset and a Multi-Task Learning based QA Framework
%U https://doi.org/10.18653%2Fv1%2F2021.findings-emnlp.392 - 1.Kuhnke, F., Ihler, S., and Ostermann, J. (2021) Relative Pose Consistency for Semi-Supervised Head Pose Estimation. In 16th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2021).
@inproceedings{KuhIhl2021,
author = {Kuhnke, Felix and Ihler, Sontje and Ostermann, Jörn},
booktitle = {16th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2021)},
keywords = {Consistency Estimation Head Pose leibnizailab myown},
month = {dec},
title = {Relative Pose Consistency for Semi-Supervised Head Pose Estimation},
year = 2021
}%0 Conference Paper
%1 KuhIhl2021
%A Kuhnke, Felix
%A Ihler, Sontje
%A Ostermann, Jörn
%B 16th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2021)
%D 2021
%T Relative Pose Consistency for Semi-Supervised Head Pose Estimation - 1.Knura, M., Kluger, F., Zahtila, M., Schiewe, J., Rosenhahn, B., and Burghardt, D. (2021) Using Object Detection on Social Media Images for Urban Bicycle Infrastructure Planning: A Case Study of Dresden, ISPRS International Journal of Geo-Information.
@article{KnuKlu2021,
author = {Knura, Martin and Kluger, Florian and Zahtila, Moris and Schiewe, Jochen and Rosenhahn, Bodo and Burghardt, Dirk},
journal = {ISPRS International Journal of Geo-Information},
keywords = {Bicycle Dresden Infrastructure Study Urban leibnizailab myown of},
month = {oct},
title = {Using Object Detection on Social Media Images for Urban Bicycle Infrastructure Planning: A Case Study of Dresden},
year = 2021
}%0 Journal Article
%1 KnuKlu2021
%A Knura, Martin
%A Kluger, Florian
%A Zahtila, Moris
%A Schiewe, Jochen
%A Rosenhahn, Bodo
%A Burghardt, Dirk
%D 2021
%J ISPRS International Journal of Geo-Information
%R 10.3390/ijgi10110733
%T Using Object Detection on Social Media Images for Urban Bicycle Infrastructure Planning: A Case Study of Dresden
%U https://doi.org/10.3390%2Fijgi10110733 - 1.Roy, S., Chakraborty, S., Mandal, A., Balde, G., Sharma, P., Natarajan, A., Khosla, M., Sural, S., and Ganguly, N. (2021) Knowledge-Aware Neural Networks for Medical Forum Question Classification. In Proceedings of the 30th ACM International Conference on Information & Knowledge Management, pp. 3398–3402, Association for Computing Machinery, New York, NY, USA.Online medical forums have become a predominant platform for answering health-related information needs of consumers. However, with a significant rise in the number of queries and the limited availability of experts, it is necessary to automatically classify medical queries based on a consumer's intention, so that these questions may be directed to the right set of medical experts. Here, we develop a novel medical knowledge-aware BERT-based model (MedBERT) that explicitly gives more weightage to medical concept-bearing words, and utilize domain-specific side information obtained from a popular medical knowledge base. We also contribute a multi-label dataset for the Medical Forum Question Classification (MFQC) task. MedBERT achieves state-of-the-art performance on two benchmark datasets and performs very well in low resource settings.
@inbook{10.1145/3459637.3482128,
abstract = {Online medical forums have become a predominant platform for answering health-related information needs of consumers. However, with a significant rise in the number of queries and the limited availability of experts, it is necessary to automatically classify medical queries based on a consumer's intention, so that these questions may be directed to the right set of medical experts. Here, we develop a novel medical knowledge-aware BERT-based model (MedBERT) that explicitly gives more weightage to medical concept-bearing words, and utilize domain-specific side information obtained from a popular medical knowledge base. We also contribute a multi-label dataset for the Medical Forum Question Classification (MFQC) task. MedBERT achieves state-of-the-art performance on two benchmark datasets and performs very well in low resource settings.},
address = {New York, NY, USA},
author = {Roy, Soumyadeep and Chakraborty, Sudip and Mandal, Aishik and Balde, Gunjan and Sharma, Prakhar and Natarajan, Anandhavelu and Khosla, Megha and Sural, Shamik and Ganguly, Niloy},
booktitle = {Proceedings of the 30th ACM International Conference on Information & Knowledge Management},
keywords = {leibnizailab myown},
pages = {3398–3402},
publisher = {Association for Computing Machinery},
title = {Knowledge-Aware Neural Networks for Medical Forum Question Classification},
year = 2021
}%0 Book Section
%1 10.1145/3459637.3482128
%A Roy, Soumyadeep
%A Chakraborty, Sudip
%A Mandal, Aishik
%A Balde, Gunjan
%A Sharma, Prakhar
%A Natarajan, Anandhavelu
%A Khosla, Megha
%A Sural, Shamik
%A Ganguly, Niloy
%B Proceedings of the 30th ACM International Conference on Information & Knowledge Management
%C New York, NY, USA
%D 2021
%I Association for Computing Machinery
%P 3398–3402
%T Knowledge-Aware Neural Networks for Medical Forum Question Classification
%U https://doi.org/10.1145/3459637.3482128
%X Online medical forums have become a predominant platform for answering health-related information needs of consumers. However, with a significant rise in the number of queries and the limited availability of experts, it is necessary to automatically classify medical queries based on a consumer's intention, so that these questions may be directed to the right set of medical experts. Here, we develop a novel medical knowledge-aware BERT-based model (MedBERT) that explicitly gives more weightage to medical concept-bearing words, and utilize domain-specific side information obtained from a popular medical knowledge base. We also contribute a multi-label dataset for the Medical Forum Question Classification (MFQC) task. MedBERT achieves state-of-the-art performance on two benchmark datasets and performs very well in low resource settings.
%@ 9781450384469 - 1.Benjamins, C., Eimer, T., Schubert, F., Biedenkapp, A., Rosenhahn, B., Hutter, F., and Lindauer, M. (2021) CARL: A Benchmark for Contextual and Adaptive Reinforcement Learning. In NeurIPS 2021 Workshop on Ecological Theory of Reinforcement Learning.
@inproceedings{BenEim2021a,
author = {Benjamins, Carolin and Eimer, Theresa and Schubert, Frederik and Biedenkapp, André and Rosenhahn, Bodo and Hutter, Frank and Lindauer, Marius},
booktitle = {NeurIPS 2021 Workshop on Ecological Theory of Reinforcement Learning},
keywords = {Adaptive Contextual Learning Reinforcement and leibnizailab myown},
month = {dec},
title = {CARL: A Benchmark for Contextual and Adaptive Reinforcement Learning},
year = 2021
}%0 Conference Paper
%1 BenEim2021a
%A Benjamins, Carolin
%A Eimer, Theresa
%A Schubert, Frederik
%A Biedenkapp, André
%A Rosenhahn, Bodo
%A Hutter, Frank
%A Lindauer, Marius
%B NeurIPS 2021 Workshop on Ecological Theory of Reinforcement Learning
%D 2021
%T CARL: A Benchmark for Contextual and Adaptive Reinforcement Learning - 1.Gritzner, D., Hinrichs, H., Stetter, C., Wielert, H., Breitner, M. H., and Ostermann, J. (2021) Wind Turbine Localization in Satellite and Aerial Images. In Proceedings of the Wind Energy Science Conference 2021, pp. 40–41.
@inproceedings{GriHin2021,
author = {Gritzner, Daniel and Hinrichs, Hauke and Stetter, Chris and Wielert, Henrik and Breitner, Michael H. and Ostermann, Jörn},
booktitle = {Proceedings of the Wind Energy Science Conference 2021},
keywords = {Aerial Images Satellite Turbine Wind leibnizailab myown},
number = 10,
pages = {40-41},
title = {Wind Turbine Localization in Satellite and Aerial Images},
year = 2021
}%0 Conference Paper
%1 GriHin2021
%A Gritzner, Daniel
%A Hinrichs, Hauke
%A Stetter, Chris
%A Wielert, Henrik
%A Breitner, Michael H.
%A Ostermann, Jörn
%B Proceedings of the Wind Energy Science Conference 2021
%D 2021
%N 10
%P 40-41
%T Wind Turbine Localization in Satellite and Aerial Images - 1.Eimer, T., Benjamins, C., and Lindauer, M. (2021) Hyperparameters in Contextual RL are Highly Situational. In NeurIPS 2021 Workshop on Ecological Theory of Reinforcement Learning.
@inproceedings{EimBen2021a,
author = {Eimer, Theresa and Benjamins, Carolin and Lindauer, Marius},
booktitle = {NeurIPS 2021 Workshop on Ecological Theory of Reinforcement Learning},
keywords = {Contextual RL leibnizailab myown},
month = {dec},
title = {Hyperparameters in Contextual RL are Highly Situational},
year = 2021
}%0 Conference Paper
%1 EimBen2021a
%A Eimer, Theresa
%A Benjamins, Carolin
%A Lindauer, Marius
%B NeurIPS 2021 Workshop on Ecological Theory of Reinforcement Learning
%D 2021
%T Hyperparameters in Contextual RL are Highly Situational - 1.Hartmann, F., and Ostermann, J. (2021) Investigation of the Effect of the Flight Path on the Three Dimensional Locatability of Targets. In Synthetic Aperture Radar (APSAR), 2021 IEEE 7th Asia-Pacific Conference.
@inproceedings{HarOst2021,
author = {Hartmann, Fabian and Ostermann, Jörn},
booktitle = {Synthetic Aperture Radar (APSAR), 2021 IEEE 7th Asia-Pacific Conference},
keywords = {Effect Flight Path leibnizailab myown},
month = {nov},
title = {Investigation of the Effect of the Flight Path on the Three Dimensional Locatability of Targets},
year = 2021
}%0 Conference Paper
%1 HarOst2021
%A Hartmann, Fabian
%A Ostermann, Jörn
%B Synthetic Aperture Radar (APSAR), 2021 IEEE 7th Asia-Pacific Conference
%D 2021
%T Investigation of the Effect of the Flight Path on the Three Dimensional Locatability of Targets - 1.Kadra, A., Lindauer, M., Hutter, F., and Grabocka, J. (2021) Regularization is all you Need: Simple Neural Nets can Excel on Tabular Data. In Proceedings of the international conference on Neural Information Processing Systems (NeurIPS).
@inproceedings{KadLin2021a,
author = {Kadra, Arlind and Lindauer, Marius and Hutter, Frank and Grabocka, Josif},
booktitle = {Proceedings of the international conference on Neural Information Processing Systems (NeurIPS)},
keywords = {Data Nets Neural Tabular leibnizailab myown},
month = {dec},
title = {Regularization is all you Need: Simple Neural Nets can Excel on Tabular Data},
year = 2021
}%0 Conference Paper
%1 KadLin2021a
%A Kadra, Arlind
%A Lindauer, Marius
%A Hutter, Frank
%A Grabocka, Josif
%B Proceedings of the international conference on Neural Information Processing Systems (NeurIPS)
%D 2021
%T Regularization is all you Need: Simple Neural Nets can Excel on Tabular Data
%U https://arxiv.org/abs/2106.11189 - 1.Mukherjee, R., Naik, A., Poddar, S., Dasgupta, S., and Ganguly, N. (2021) Understanding the Role of Affect Dimensions in Detecting Emotions from Tweets: A Multi-task Approach. In SIGIR 2021.We propose VADEC, a multi-task framework that exploits the correlation between the categorical and dimensional models of emotion representation for better subjectivity analysis. Focusing primarily on the effective detection of emotions from tweets, we jointly train multi-label emotion classification and multi-dimensional emotion regression, thereby utilizing the inter-relatedness between the tasks. Co-training especially helps in improving the performance of the classification task as we outperform the strongest baselines with 3.4%, 11%, and 3.9% gains in Jaccard Accuracy, Macro-F1, and Micro-F1 scores respectively on the AIT dataset. We also achieve state-of-the-art results with 11.3% gains averaged over six different metrics on the SenWave dataset. For the regression task, VADEC, when trained with SenWave, achieves 7.6% and 16.5% gains in Pearson Correlation scores over the current state-of-the-art on the EMOBANK dataset for the Valence (V) and Dominance (D) affect dimensions respectively. We conclude our work with a case study on COVID-19 tweets posted by Indians that further helps in establishing the efficacy of our proposed solution.
@inproceedings{mukherjee2021understanding,
abstract = {We propose VADEC, a multi-task framework that exploits the correlation between the categorical and dimensional models of emotion representation for better subjectivity analysis. Focusing primarily on the effective detection of emotions from tweets, we jointly train multi-label emotion classification and multi-dimensional emotion regression, thereby utilizing the inter-relatedness between the tasks. Co-training especially helps in improving the performance of the classification task as we outperform the strongest baselines with 3.4%, 11%, and 3.9% gains in Jaccard Accuracy, Macro-F1, and Micro-F1 scores respectively on the AIT dataset. We also achieve state-of-the-art results with 11.3% gains averaged over six different metrics on the SenWave dataset. For the regression task, VADEC, when trained with SenWave, achieves 7.6% and 16.5% gains in Pearson Correlation scores over the current state-of-the-art on the EMOBANK dataset for the Valence (V) and Dominance (D) affect dimensions respectively. We conclude our work with a case study on COVID-19 tweets posted by Indians that further helps in establishing the efficacy of our proposed solution.},
author = {Mukherjee, Rajdeep and Naik, Atharva and Poddar, Sriyash and Dasgupta, Soham and Ganguly, Niloy},
booktitle = {SIGIR 2021},
keywords = {leibnizailab myown},
title = {Understanding the Role of Affect Dimensions in Detecting Emotions from Tweets: A Multi-task Approach},
year = 2021
}%0 Conference Paper
%1 mukherjee2021understanding
%A Mukherjee, Rajdeep
%A Naik, Atharva
%A Poddar, Sriyash
%A Dasgupta, Soham
%A Ganguly, Niloy
%B SIGIR 2021
%D 2021
%R 10.1145/3404835.3463080
%T Understanding the Role of Affect Dimensions in Detecting Emotions from Tweets: A Multi-task Approach
%U http://arxiv.org/abs/2105.03983
%X We propose VADEC, a multi-task framework that exploits the correlation between the categorical and dimensional models of emotion representation for better subjectivity analysis. Focusing primarily on the effective detection of emotions from tweets, we jointly train multi-label emotion classification and multi-dimensional emotion regression, thereby utilizing the inter-relatedness between the tasks. Co-training especially helps in improving the performance of the classification task as we outperform the strongest baselines with 3.4%, 11%, and 3.9% gains in Jaccard Accuracy, Macro-F1, and Micro-F1 scores respectively on the AIT dataset. We also achieve state-of-the-art results with 11.3% gains averaged over six different metrics on the SenWave dataset. For the regression task, VADEC, when trained with SenWave, achieves 7.6% and 16.5% gains in Pearson Correlation scores over the current state-of-the-art on the EMOBANK dataset for the Valence (V) and Dominance (D) affect dimensions respectively. We conclude our work with a case study on COVID-19 tweets posted by Indians that further helps in establishing the efficacy of our proposed solution. - 1.Zimmer, L., Lindauer, M., and Hutter, F. (2021) Auto-PyTorch Tabular: Multi-Fidelity MetaLearning for Efficient and Robust AutoDL, IEEE Transactions on Pattern Analysis and Machine Intelligence 43, 3079–3090.
@article{ZimLin2021a,
author = {Zimmer, Lucas and Lindauer, Marius and Hutter, Frank},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
keywords = {Auto-PyTorch MetaLearning leibnizailab myown},
month = {aug},
number = 9,
pages = {3079 - 3090},
title = {Auto-PyTorch Tabular: Multi-Fidelity MetaLearning for Efficient and Robust AutoDL},
volume = 43,
year = 2021
}%0 Journal Article
%1 ZimLin2021a
%A Zimmer, Lucas
%A Lindauer, Marius
%A Hutter, Frank
%D 2021
%J IEEE Transactions on Pattern Analysis and Machine Intelligence
%N 9
%P 3079 - 3090
%T Auto-PyTorch Tabular: Multi-Fidelity MetaLearning for Efficient and Robust AutoDL
%U /brokenurl#Arxiv, IEEE TPAMI
%V 43 - 1.Pestel-Schiller, U., and Ostermann, J. (2021) Interpreter-Based Evaluation of Compressed SAR Images Using JPEG and HEVC Intra Coding: Compression Can Improve Usability. In 13th European Conference on Synthetic Aperture Radar.
@inproceedings{PesOst2021a,
author = {Pestel-Schiller, Ulrike and Ostermann, Jörn},
booktitle = {13th European Conference on Synthetic Aperture Radar},
keywords = {Compressed HEVC Images JPEG SAR and leibnizailab myown},
month = {mar},
title = {Interpreter-Based Evaluation of Compressed SAR Images Using JPEG and HEVC Intra Coding: Compression Can Improve Usability},
year = 2021
}%0 Conference Paper
%1 PesOst2021a
%A Pestel-Schiller, Ulrike
%A Ostermann, Jörn
%B 13th European Conference on Synthetic Aperture Radar
%D 2021
%T Interpreter-Based Evaluation of Compressed SAR Images Using JPEG and HEVC Intra Coding: Compression Can Improve Usability - 1.Truong, G., Le, H., Suter, D., Zhang, E., and Gilani, S. Z. (2021) Unsupervised Learning for Robust Fitting: A Reinforcement Learning Approach. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10343–10352.
@inproceedings{giang_cvpr2021,
author = {Truong, Giang and Le, Huu and Suter, David and Zhang, Erchuan and Gilani, Syed Zulqarnain},
booktitle = {2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
keywords = {leibnizailab myown},
pages = {10343-10352},
title = {Unsupervised Learning for Robust Fitting: A Reinforcement Learning Approach},
year = 2021
}%0 Conference Paper
%1 giang_cvpr2021
%A Truong, Giang
%A Le, Huu
%A Suter, David
%A Zhang, Erchuan
%A Gilani, Syed Zulqarnain
%B 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
%D 2021
%P 10343-10352
%R 10.1109/CVPR46437.2021.01021
%T Unsupervised Learning for Robust Fitting: A Reinforcement Learning Approach - 1.Hartmann, F., Sommer, A., Pestel-Schiller, U., and Osterman, J. (2021) A scheme for stabilizing the image generation for VideoSAR. In 13th European Conference on Synthetic Aperture Radar.
@inproceedings{HarSom2021a,
author = {Hartmann, Fabian and Sommer, Aron and Pestel-Schiller, Ulrike and Osterman, Jörn},
booktitle = {13th European Conference on Synthetic Aperture Radar},
keywords = {VideoSAR leibnizailab myown},
month = {mar},
title = {A scheme for stabilizing the image generation for VideoSAR},
year = 2021
}%0 Conference Paper
%1 HarSom2021a
%A Hartmann, Fabian
%A Sommer, Aron
%A Pestel-Schiller, Ulrike
%A Osterman, Jörn
%B 13th European Conference on Synthetic Aperture Radar
%D 2021
%T A scheme for stabilizing the image generation for VideoSAR - 1.Rudolph, M., Wandt, B., and Rosenhahn, B. (2021) Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows. In Winter Conference on Applications of Computer Vision (WACV).
@inproceedings{RudWan2021a,
author = {Rudolph, Marco and Wandt, Bastian and Rosenhahn, Bodo},
booktitle = {Winter Conference on Applications of Computer Vision (WACV)},
keywords = {Defect Detection Semi-Supervised leibnizailab myown},
month = {jan},
title = {Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows},
year = 2021
}%0 Conference Paper
%1 RudWan2021a
%A Rudolph, Marco
%A Wandt, Bastian
%A Rosenhahn, Bodo
%B Winter Conference on Applications of Computer Vision (WACV)
%D 2021
%T Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows
%U /brokenurl#arxiv, GitHub, YouTube - 1.Roy, S., Sural, S., Chhaya, N., Natarajan, A., and Ganguly, N. (2021) An Integrated Approach for Improving Brand Consistency of Web Content: Modeling, Analysis and Recommendation., ACM Trans. Web 15, 9:1–9:25.
@article{journals/corr/abs-2011-09754,
author = {Roy, Soumyadeep and Sural, Shamik and Chhaya, Niyati and Natarajan, Anandhavelu and Ganguly, Niloy},
journal = {ACM Trans. Web},
keywords = {leibnizailab},
number = 2,
pages = {9:1-9:25},
title = {An Integrated Approach for Improving Brand Consistency of Web Content: Modeling, Analysis and Recommendation.},
volume = 15,
year = 2021
}%0 Journal Article
%1 journals/corr/abs-2011-09754
%A Roy, Soumyadeep
%A Sural, Shamik
%A Chhaya, Niyati
%A Natarajan, Anandhavelu
%A Ganguly, Niloy
%D 2021
%J ACM Trans. Web
%N 2
%P 9:1-9:25
%T An Integrated Approach for Improving Brand Consistency of Web Content: Modeling, Analysis and Recommendation.
%U http://dblp.uni-trier.de/db/journals/corr/corr2011.html#abs-2011-09754
%V 15 - 1.Kabongo, S., D’Souza, J., and Auer, S. (2021) Automated Mining of Leaderboards for Empirical AI Research, springer, International Conference on Asian Digital Libraries ICADL 2021: Towards Open and Trustworthy Digital Societies, 453–470.With the rapid growth of research publications, empowering scientists to keep an oversight over scientific progress is of paramount importance. In this regard, the leaderboards facet of information organization provides an overview on the state-of-the-art by aggregating empirical results from various studies addressing the same research challenge. Crowdsourcing efforts like PAPERSWITHCODE among others are devoted to the construction of leaderboards predominantly for various subdomains in Artificial Intelligence. Leaderboards provide machine-readable scholarly knowledge that has proven to be directly useful for scientists to keep track of research progress – their construction could be greatly expedited with automated text mining. This study presents a comprehensive approach for generating leaderboards for knowledge-graph-based scholarly information organization. Specifically, we investigate the problem of automated leaderboard construction using state-of-the-art transformer models, viz. Bert, SciBert, and XLNet. Our analysis reveals an optimal approach that significantly outperforms existing baselines for the task with evaluation scores above 90% in F1. This, in turn, offers new state-of-the-art results for leaderboard extraction. As a result, a vast share of empirical AI research can be organized in the next-generation digital libraries as knowledge graphs.
@article{DBLP:conf/icadl/KabongoDA21,
abstract = {With the rapid growth of research publications, empowering scientists to keep an oversight over scientific progress is of paramount importance. In this regard, the leaderboards facet of information organization provides an overview on the state-of-the-art by aggregating empirical results from various studies addressing the same research challenge. Crowdsourcing efforts like PAPERSWITHCODE among others are devoted to the construction of leaderboards predominantly for various subdomains in Artificial Intelligence. Leaderboards provide machine-readable scholarly knowledge that has proven to be directly useful for scientists to keep track of research progress – their construction could be greatly expedited with automated text mining. This study presents a comprehensive approach for generating leaderboards for knowledge-graph-based scholarly information organization. Specifically, we investigate the problem of automated leaderboard construction using state-of-the-art transformer models, viz. Bert, SciBert, and XLNet. Our analysis reveals an optimal approach that significantly outperforms existing baselines for the task with evaluation scores above 90% in F1. This, in turn, offers new state-of-the-art results for leaderboard extraction. As a result, a vast share of empirical AI research can be organized in the next-generation digital libraries as knowledge graphs.},
author = {Kabongo, Salomon and D'Souza, Jennifer and Auer, Sören},
journal = {springer, International Conference on Asian Digital Libraries},
keywords = {Information-extraction Knowledge-graphs Neural-machine-learning Scholarly-text-mining Table-mining leibnizailab myown},
month = {November 2021},
pages = {453–470},
title = {Automated Mining of Leaderboards for Empirical AI Research},
volume = {ICADL 2021: Towards Open and Trustworthy Digital Societies},
year = 2021
}%0 Journal Article
%1 DBLP:conf/icadl/KabongoDA21
%A Kabongo, Salomon
%A D'Souza, Jennifer
%A Auer, Sören
%D 2021
%J springer, International Conference on Asian Digital Libraries
%P 453–470
%R https://doi.org/10.1007/978-3-030-91669-5_35
%T Automated Mining of Leaderboards for Empirical AI Research
%U https://link.springer.com/chapter/10.1007/978-3-030-91669-5_35
%V ICADL 2021: Towards Open and Trustworthy Digital Societies
%X With the rapid growth of research publications, empowering scientists to keep an oversight over scientific progress is of paramount importance. In this regard, the leaderboards facet of information organization provides an overview on the state-of-the-art by aggregating empirical results from various studies addressing the same research challenge. Crowdsourcing efforts like PAPERSWITHCODE among others are devoted to the construction of leaderboards predominantly for various subdomains in Artificial Intelligence. Leaderboards provide machine-readable scholarly knowledge that has proven to be directly useful for scientists to keep track of research progress – their construction could be greatly expedited with automated text mining. This study presents a comprehensive approach for generating leaderboards for knowledge-graph-based scholarly information organization. Specifically, we investigate the problem of automated leaderboard construction using state-of-the-art transformer models, viz. Bert, SciBert, and XLNet. Our analysis reveals an optimal approach that significantly outperforms existing baselines for the task with evaluation scores above 90% in F1. This, in turn, offers new state-of-the-art results for leaderboard extraction. As a result, a vast share of empirical AI research can be organized in the next-generation digital libraries as knowledge graphs. - 1.Tan, D. W., Gilani, S. Z., Boutrus, M., Alvares, G. A., Whitehouse, A. J., Mian, A., Suter, D., and Maybery, M. T. (2021) Facial asymmetry in parents of children on the autism spectrum, Autism Research.
@article{autism:2021,
author = {Tan, D W and Gilani, S Z and Boutrus, M and Alvares, G A. and Whitehouse, A J.O. and Mian, A and Suter, D and Maybery, M T.},
journal = {Autism Research},
keywords = {leibnizailab myown},
title = {Facial asymmetry in parents of children on the autism spectrum},
year = 2021
}%0 Journal Article
%1 autism:2021
%A Tan, D W
%A Gilani, S Z
%A Boutrus, M
%A Alvares, G A.
%A Whitehouse, A J.O.
%A Mian, A
%A Suter, D
%A Maybery, M T.
%D 2021
%J Autism Research
%R 10.1002/aur.2612
%T Facial asymmetry in parents of children on the autism spectrum - 1.Chin, T.-J., Suter, D., Ch’ng, S.-F., and Quach, J. (2021) Quantum Robust Fitting. In Computer Vision -- ACCV 2020 (Ishikawa, H., Liu, C.-L., Pajdla, T., and Shi, J., Eds.), pp. 485–499, Springer International Publishing, Cham.Many computer vision applications need to recover structure from imperfect measurements of the real world. The task is often solved by robustly fitting a geometric model onto noisy and outlier-contaminated data. However, recent theoretical analyses indicate that many commonly used formulations of robust fitting in computer vision are not amenable to tractable solution and approximation. In this paper, we explore the usage of quantum computers for robust fitting. To do so, we examine and establish the practical usefulness of a robust fitting formulation inspired by the analysis of monotone Boolean functions. We then investigate a quantum algorithm to solve the formulation and analyse the computational speed-up possible over the classical algorithm. Our work thus proposes one of the first quantum treatments of robust fitting for computer vision.
@inproceedings{10.1007/978-3-030-69525-5_29,
abstract = {Many computer vision applications need to recover structure from imperfect measurements of the real world. The task is often solved by robustly fitting a geometric model onto noisy and outlier-contaminated data. However, recent theoretical analyses indicate that many commonly used formulations of robust fitting in computer vision are not amenable to tractable solution and approximation. In this paper, we explore the usage of quantum computers for robust fitting. To do so, we examine and establish the practical usefulness of a robust fitting formulation inspired by the analysis of monotone Boolean functions. We then investigate a quantum algorithm to solve the formulation and analyse the computational speed-up possible over the classical algorithm. Our work thus proposes one of the first quantum treatments of robust fitting for computer vision.},
address = {Cham},
author = {Chin, Tat-Jun and Suter, David and Ch'ng, Shin-Fang and Quach, James},
booktitle = {Computer Vision -- ACCV 2020},
editor = {Ishikawa, Hiroshi and Liu, Cheng-Lin and Pajdla, Tomas and Shi, Jianbo},
keywords = {leibnizailab myown},
pages = {485--499},
publisher = {Springer International Publishing},
title = {Quantum Robust Fitting},
year = 2021
}%0 Conference Paper
%1 10.1007/978-3-030-69525-5_29
%A Chin, Tat-Jun
%A Suter, David
%A Ch'ng, Shin-Fang
%A Quach, James
%B Computer Vision -- ACCV 2020
%C Cham
%D 2021
%E Ishikawa, Hiroshi
%E Liu, Cheng-Lin
%E Pajdla, Tomas
%E Shi, Jianbo
%I Springer International Publishing
%P 485--499
%R 10.1007/978-3-030-69525-5_29
%T Quantum Robust Fitting
%X Many computer vision applications need to recover structure from imperfect measurements of the real world. The task is often solved by robustly fitting a geometric model onto noisy and outlier-contaminated data. However, recent theoretical analyses indicate that many commonly used formulations of robust fitting in computer vision are not amenable to tractable solution and approximation. In this paper, we explore the usage of quantum computers for robust fitting. To do so, we examine and establish the practical usefulness of a robust fitting formulation inspired by the analysis of monotone Boolean functions. We then investigate a quantum algorithm to solve the formulation and analyse the computational speed-up possible over the classical algorithm. Our work thus proposes one of the first quantum treatments of robust fitting for computer vision.
%@ 978-3-030-69525-5 - 1.Sheshadri, S., Saha, A., Patel, P., Datta, S., and Ganguly, N. (2021) Graph-based semi-supervised learning through the lens of safety. In Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence (de Campos, C., and Maathuis, M. H., Eds.), pp. 1576–1586, PMLR.Graph-based semi-supervised learning (G-SSL) algorithms have witnessed rapid development and widespread usage across a variety of applications in recent years. However, the theoretical characterisation of the efficacy of such algorithms has remained an under-explored area. We introduce a novel algorithm for G-SSL, CSX, whose objective function extends those of Label Propagation and Expander, two popular G-SSL algorithms. We provide data-dependent generalisation error bounds for all three aforementioned algorithms when they are applied to graphs drawn from a partially labelled extension of a versatile latent space graph generative model. The bounds we obtain enable us to characterise the predictive performance as measured by accuracy in terms of homophily and label quantity. Building on this we develop a key notion of GLM-safety which enables us to compare G-SSL algorithms on the basis of the range of graphs on which they obtain a guaranteed accuracy. We show that the proposed algorithm CSX has a better GLM-safety profile than Label Propagation and Expander while achieving comparable or better accuracy on synthetic as well as real-world benchmark networks.
@inproceedings{pmlr-v161-sheshadri21a,
abstract = {Graph-based semi-supervised learning (G-SSL) algorithms have witnessed rapid development and widespread usage across a variety of applications in recent years. However, the theoretical characterisation of the efficacy of such algorithms has remained an under-explored area. We introduce a novel algorithm for G-SSL, CSX, whose objective function extends those of Label Propagation and Expander, two popular G-SSL algorithms. We provide data-dependent generalisation error bounds for all three aforementioned algorithms when they are applied to graphs drawn from a partially labelled extension of a versatile latent space graph generative model. The bounds we obtain enable us to characterise the predictive performance as measured by accuracy in terms of homophily and label quantity. Building on this we develop a key notion of GLM-safety which enables us to compare G-SSL algorithms on the basis of the range of graphs on which they obtain a guaranteed accuracy. We show that the proposed algorithm CSX has a better GLM-safety profile than Label Propagation and Expander while achieving comparable or better accuracy on synthetic as well as real-world benchmark networks.},
author = {Sheshadri, Shreyas and Saha, Avirup and Patel, Priyank and Datta, Samik and Ganguly, Niloy},
booktitle = {Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence},
editor = {de Campos, Cassio and Maathuis, Marloes H.},
keywords = {leibnizailab myown},
month = {27--30 Jul},
pages = {1576--1586},
publisher = {PMLR},
series = {Proceedings of Machine Learning Research},
title = {Graph-based semi-supervised learning through the lens of safety},
volume = 161,
year = 2021
}%0 Conference Paper
%1 pmlr-v161-sheshadri21a
%A Sheshadri, Shreyas
%A Saha, Avirup
%A Patel, Priyank
%A Datta, Samik
%A Ganguly, Niloy
%B Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence
%D 2021
%E de Campos, Cassio
%E Maathuis, Marloes H.
%I PMLR
%P 1576--1586
%T Graph-based semi-supervised learning through the lens of safety
%U https://proceedings.mlr.press/v161/sheshadri21a.html
%V 161
%X Graph-based semi-supervised learning (G-SSL) algorithms have witnessed rapid development and widespread usage across a variety of applications in recent years. However, the theoretical characterisation of the efficacy of such algorithms has remained an under-explored area. We introduce a novel algorithm for G-SSL, CSX, whose objective function extends those of Label Propagation and Expander, two popular G-SSL algorithms. We provide data-dependent generalisation error bounds for all three aforementioned algorithms when they are applied to graphs drawn from a partially labelled extension of a versatile latent space graph generative model. The bounds we obtain enable us to characterise the predictive performance as measured by accuracy in terms of homophily and label quantity. Building on this we develop a key notion of GLM-safety which enables us to compare G-SSL algorithms on the basis of the range of graphs on which they obtain a guaranteed accuracy. We show that the proposed algorithm CSX has a better GLM-safety profile than Label Propagation and Expander while achieving comparable or better accuracy on synthetic as well as real-world benchmark networks. - 1.Moosbauer, J., Herbinger, J., Casalicchio, G., Lindauer, M., and Bischl, B. (2021) Explaining Hyperparameter Optimization via Partial Dependence Plots. In Proceedings of the international conference on Neural Information Processing Systems (NeurIPS).
@inproceedings{MooHer2021a,
author = {Moosbauer, Julia and Herbinger, Julia and Casalicchio, Giuseppe and Lindauer, Marius and Bischl, Bernd},
booktitle = {Proceedings of the international conference on Neural Information Processing Systems (NeurIPS)},
keywords = {Dependence Hyperparameter Partial Plots leibnizailab myown},
month = {dec},
title = {Explaining Hyperparameter Optimization via Partial Dependence Plots},
year = 2021
}%0 Conference Paper
%1 MooHer2021a
%A Moosbauer, Julia
%A Herbinger, Julia
%A Casalicchio, Giuseppe
%A Lindauer, Marius
%A Bischl, Bernd
%B Proceedings of the international conference on Neural Information Processing Systems (NeurIPS)
%D 2021
%T Explaining Hyperparameter Optimization via Partial Dependence Plots - 1.Tennakoon, R., Suter, D., Zhang, E., Chin, T.-J., and Bab-Hadiashar, A. (2021) Consensus Maximisation Using Influences of Monotone Boolean Functions. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2865–2874.
@inproceedings{BMFcvpr2021,
author = {Tennakoon, Ruwan and Suter, David and Zhang, Erchuan and Chin, Tat-Jun and Bab-Hadiashar, Alireza},
booktitle = {2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
keywords = {leibnizailab myown},
note = {Oral Presentation (17\% of accepted papers, roughly 3\% of sumitted papers)},
pages = {2865-2874},
title = {Consensus Maximisation Using Influences of Monotone Boolean Functions},
year = 2021
}%0 Conference Paper
%1 BMFcvpr2021
%A Tennakoon, Ruwan
%A Suter, David
%A Zhang, Erchuan
%A Chin, Tat-Jun
%A Bab-Hadiashar, Alireza
%B 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
%D 2021
%P 2865-2874
%R 10.1109/CVPR46437.2021.00289
%T Consensus Maximisation Using Influences of Monotone Boolean Functions - 1.Ilyas, Z., Sharif, N., Schousboe, J. T., Lewis, J. R., Suter, D., and Gilani, S. Z. (2021) GuideNet: Learning Inter- Vertebral Guides in DXA Lateral Spine Images. In 2021 Digital Image Computing: Techniques and Applications (DICTA), pp. 01–07.
@inproceedings{9647067,
author = {Ilyas, Zaid and Sharif, Naeha and Schousboe, John T. and Lewis, Joshua R. and Suter, David and Gilani, Syed Zulqarnain},
booktitle = {2021 Digital Image Computing: Techniques and Applications (DICTA)},
keywords = {leibnizailab myown},
pages = {01-07},
title = {GuideNet: Learning Inter- Vertebral Guides in DXA Lateral Spine Images},
year = 2021
}%0 Conference Paper
%1 9647067
%A Ilyas, Zaid
%A Sharif, Naeha
%A Schousboe, John T.
%A Lewis, Joshua R.
%A Suter, David
%A Gilani, Syed Zulqarnain
%B 2021 Digital Image Computing: Techniques and Applications (DICTA)
%D 2021
%P 01-07
%R 10.1109/DICTA52665.2021.9647067
%T GuideNet: Learning Inter- Vertebral Guides in DXA Lateral Spine Images - 1.Das, S., Patibandla, H., Bhattacharya, S., Bera, K., Ganguly, N., and Bhattacharya, S. (2021) TMCOSS: Thresholded Multi-Criteria Online Subset Selection for Data-Efficient Autonomous Driving. In ICCV.
@inproceedings{das2021tmcoss,
author = {Das, Soumi and Patibandla, Harikrishna and Bhattacharya, Suparna and Bera, Kshounis and Ganguly, Niloy and Bhattacharya, Sourangshu},
booktitle = {ICCV},
keywords = {leibnizailab myown},
title = {TMCOSS: Thresholded Multi-Criteria Online Subset Selection for Data-Efficient Autonomous Driving},
year = 2021
}%0 Conference Paper
%1 das2021tmcoss
%A Das, Soumi
%A Patibandla, Harikrishna
%A Bhattacharya, Suparna
%A Bera, Kshounis
%A Ganguly, Niloy
%A Bhattacharya, Sourangshu
%B ICCV
%D 2021
%T TMCOSS: Thresholded Multi-Criteria Online Subset Selection for Data-Efficient Autonomous Driving - 1.Ghosh, S., Ganguly, N., Mitra, B., and De, P. (2021) Designing an Experience Sampling Method for Smartphone Based Emotion Detection, IEEE Transactions on Affective Computing 12, 913–927.Smartphones provide the capability to perform in-situ sampling of human behavior using Experience Sampling Method (ESM). Designing an ESM schedule involves probing the user repeatedly at suitable moments to collect self-reports. Timely probe generation to collect high fidelity user responses while keeping probing rate low is challenging. In mobile-based ESM, timeliness of the probe is also impacted by user's availability to respond to self-report request. Thus, a good ESM design must consider -
probing frequency ,timely self-report collection , andnotifying at opportune moment to ensure highresponse quality . We propose a two-phase ESM design, where the first phase (a) balances between probing frequency and self-report timeliness, and (b) in parallel, constructs a predictive model to identify opportune probing moments. The second phase uses this model to further improve response quality by eliminating inopportune probes. We use typing-based emotion detection in smartphone as a case study to validate proposed ESM design. Our results demonstrate that it reduces probing rate by 64 percent, samples self-reports timely by reducing elapsed time between self-report collection, and event trigger by 9 percent while detecting inopportune moments with an average accuracy of 89 percent. These design choices improve the response quality, as manifested by 96 percent valid response collection and a maximum improvement of 24 percent in emotion classification accuracy.@article{8668435,
abstract = {Smartphones provide the capability to perform in-situ sampling of human behavior using Experience Sampling Method (ESM). Designing an ESM schedule involves probing the user repeatedly at suitable moments to collect self-reports. Timely probe generation to collect high fidelity user responses while keeping probing rate low is challenging. In mobile-based ESM, timeliness of the probe is also impacted by user's availability to respond to self-report request. Thus, a good ESM design must consider -probing frequency ,timely self-report collection , andnotifying at opportune moment to ensure highresponse quality . We propose a two-phase ESM design, where the first phase (a) balances between probing frequency and self-report timeliness, and (b) in parallel, constructs a predictive model to identify opportune probing moments. The second phase uses this model to further improve response quality by eliminating inopportune probes. We use typing-based emotion detection in smartphone as a case study to validate proposed ESM design. Our results demonstrate that it reduces probing rate by 64 percent, samples self-reports timely by reducing elapsed time between self-report collection, and event trigger by 9 percent while detecting inopportune moments with an average accuracy of 89 percent. These design choices improve the response quality, as manifested by 96 percent valid response collection and a maximum improvement of 24 percent in emotion classification accuracy.},
author = {Ghosh, Surjya and Ganguly, Niloy and Mitra, Bivas and De, Pradipta},
journal = {IEEE Transactions on Affective Computing},
keywords = {leibnizailab myown},
month = {oct},
number = 4,
pages = {913-927},
title = {Designing an Experience Sampling Method for Smartphone Based Emotion Detection},
volume = 12,
year = 2021
}%0 Journal Article
%1 8668435
%A Ghosh, Surjya
%A Ganguly, Niloy
%A Mitra, Bivas
%A De, Pradipta
%D 2021
%J IEEE Transactions on Affective Computing
%N 4
%P 913-927
%R 10.1109/TAFFC.2019.2905561
%T Designing an Experience Sampling Method for Smartphone Based Emotion Detection
%V 12
%X Smartphones provide the capability to perform in-situ sampling of human behavior using Experience Sampling Method (ESM). Designing an ESM schedule involves probing the user repeatedly at suitable moments to collect self-reports. Timely probe generation to collect high fidelity user responses while keeping probing rate low is challenging. In mobile-based ESM, timeliness of the probe is also impacted by user's availability to respond to self-report request. Thus, a good ESM design must consider -probing frequency ,timely self-report collection , andnotifying at opportune moment to ensure highresponse quality . We propose a two-phase ESM design, where the first phase (a) balances between probing frequency and self-report timeliness, and (b) in parallel, constructs a predictive model to identify opportune probing moments. The second phase uses this model to further improve response quality by eliminating inopportune probes. We use typing-based emotion detection in smartphone as a case study to validate proposed ESM design. Our results demonstrate that it reduces probing rate by 64 percent, samples self-reports timely by reducing elapsed time between self-report collection, and event trigger by 9 percent while detecting inopportune moments with an average accuracy of 89 percent. These design choices improve the response quality, as manifested by 96 percent valid response collection and a maximum improvement of 24 percent in emotion classification accuracy. - 1.Koley, P., Saha, A., Bhattacharya, S., Ganguly, N., and De, A. (2021) Demarcating Endogenous and Exogenous Opinion Dynamics: An Experimental Design Approach, ACM Trans. Knowl. Discov. Data, Association for Computing Machinery, New York, NY, USA 15.The networked opinion diffusion in online social networks is often governed by the two genres of opinions—endogenous opinions that are driven by the influence of social contacts among users, and exogenous opinions which are formed by external effects like news and feeds. Accurate demarcation of endogenous and exogenous messages offers an important cue to opinion modeling, thereby enhancing its predictive performance. In this article, we design a suite of unsupervised classification methods based on experimental design approaches, in which, we aim to select the subsets of events which minimize different measures of mean estimation error. In more detail, we first show that these subset selection tasks are NP-Hard. Then we show that the associated objective functions are weakly submodular, which allows us to cast efficient approximation algorithms with guarantees. Finally, we validate the efficacy of our proposal on various real-world datasets crawled from Twitter as well as diverse synthetic datasets. Our experiments range from validating prediction performance on unsanitized and sanitized events to checking the effect of selecting optimal subsets of various sizes. Through various experiments, we have found that our method offers a significant improvement in accuracy in terms of opinion forecasting, against several competitors.
@article{10.1145/3449361,
abstract = {The networked opinion diffusion in online social networks is often governed by the two genres of opinions—endogenous opinions that are driven by the influence of social contacts among users, and exogenous opinions which are formed by external effects like news and feeds. Accurate demarcation of endogenous and exogenous messages offers an important cue to opinion modeling, thereby enhancing its predictive performance. In this article, we design a suite of unsupervised classification methods based on experimental design approaches, in which, we aim to select the subsets of events which minimize different measures of mean estimation error. In more detail, we first show that these subset selection tasks are NP-Hard. Then we show that the associated objective functions are weakly submodular, which allows us to cast efficient approximation algorithms with guarantees. Finally, we validate the efficacy of our proposal on various real-world datasets crawled from Twitter as well as diverse synthetic datasets. Our experiments range from validating prediction performance on unsanitized and sanitized events to checking the effect of selecting optimal subsets of various sizes. Through various experiments, we have found that our method offers a significant improvement in accuracy in terms of opinion forecasting, against several competitors.},
address = {New York, NY, USA},
author = {Koley, Paramita and Saha, Avirup and Bhattacharya, Sourangshu and Ganguly, Niloy and De, Abir},
journal = {ACM Trans. Knowl. Discov. Data},
keywords = {leibnizailab myown},
month = {jun},
number = 6,
publisher = {Association for Computing Machinery},
title = {Demarcating Endogenous and Exogenous Opinion Dynamics: An Experimental Design Approach},
volume = 15,
year = 2021
}%0 Journal Article
%1 10.1145/3449361
%A Koley, Paramita
%A Saha, Avirup
%A Bhattacharya, Sourangshu
%A Ganguly, Niloy
%A De, Abir
%C New York, NY, USA
%D 2021
%I Association for Computing Machinery
%J ACM Trans. Knowl. Discov. Data
%N 6
%R 10.1145/3449361
%T Demarcating Endogenous and Exogenous Opinion Dynamics: An Experimental Design Approach
%U https://doi.org/10.1145/3449361
%V 15
%X The networked opinion diffusion in online social networks is often governed by the two genres of opinions—endogenous opinions that are driven by the influence of social contacts among users, and exogenous opinions which are formed by external effects like news and feeds. Accurate demarcation of endogenous and exogenous messages offers an important cue to opinion modeling, thereby enhancing its predictive performance. In this article, we design a suite of unsupervised classification methods based on experimental design approaches, in which, we aim to select the subsets of events which minimize different measures of mean estimation error. In more detail, we first show that these subset selection tasks are NP-Hard. Then we show that the associated objective functions are weakly submodular, which allows us to cast efficient approximation algorithms with guarantees. Finally, we validate the efficacy of our proposal on various real-world datasets crawled from Twitter as well as diverse synthetic datasets. Our experiments range from validating prediction performance on unsanitized and sanitized events to checking the effect of selecting optimal subsets of various sizes. Through various experiments, we have found that our method offers a significant improvement in accuracy in terms of opinion forecasting, against several competitors. - 1.Samanta, B., Agrawal, M., and Ganguly, N. (2021) A Hierarchical VAE for Calibrating Attributes while Generating Text using Normalizing Flow, pp. 2405–2415, Association for Computational Linguistics.In this digital age, online users expect personalized content. To cater to diverse group of audiences across online platforms it is necessary to generate multiple variants of same content with differing degree of characteristics (sentiment, style, formality, etc.). Though text-style transfer is a well explored related area, it focuses on flipping the style attribute polarity instead of regulating a fine-grained attribute transfer. In this paper we propose a hierarchical architecture for finer control over the at- tribute, preserving content using attribute dis- entanglement. We demonstrate the effective- ness of the generative process for two different attributes with varied complexity, namely sentiment and formality. With extensive experiments and human evaluation on five real-world datasets, we show that the framework can generate natural looking sentences with finer degree of control of intensity of a given attribute.
@proceedings{samanta2021hierarchical,
abstract = {In this digital age, online users expect personalized content. To cater to diverse group of audiences across online platforms it is necessary to generate multiple variants of same content with differing degree of characteristics (sentiment, style, formality, etc.). Though text-style transfer is a well explored related area, it focuses on flipping the style attribute polarity instead of regulating a fine-grained attribute transfer. In this paper we propose a hierarchical architecture for finer control over the at- tribute, preserving content using attribute dis- entanglement. We demonstrate the effective- ness of the generative process for two different attributes with varied complexity, namely sentiment and formality. With extensive experiments and human evaluation on five real-world datasets, we show that the framework can generate natural looking sentences with finer degree of control of intensity of a given attribute.},
address = {Association for Computational Linguistics},
author = {Samanta, Bidisha and Agrawal, Mohit and Ganguly, NIloy},
howpublished = {Online},
keywords = {leibnizailab myown},
pages = {2405-2415},
title = {A Hierarchical VAE for Calibrating Attributes while Generating Text using Normalizing Flow},
year = 2021
}%0 Conference Proceedings
%1 samanta2021hierarchical
%A Samanta, Bidisha
%A Agrawal, Mohit
%A Ganguly, NIloy
%C Association for Computational Linguistics
%D 2021
%P 2405-2415
%T A Hierarchical VAE for Calibrating Attributes while Generating Text using Normalizing Flow
%U https://aclanthology.org/2021.acl-long.187
%X In this digital age, online users expect personalized content. To cater to diverse group of audiences across online platforms it is necessary to generate multiple variants of same content with differing degree of characteristics (sentiment, style, formality, etc.). Though text-style transfer is a well explored related area, it focuses on flipping the style attribute polarity instead of regulating a fine-grained attribute transfer. In this paper we propose a hierarchical architecture for finer control over the at- tribute, preserving content using attribute dis- entanglement. We demonstrate the effective- ness of the generative process for two different attributes with varied complexity, namely sentiment and formality. With extensive experiments and human evaluation on five real-world datasets, we show that the framework can generate natural looking sentences with finer degree of control of intensity of a given attribute. - 1.Mukherjee, A., Mallick, M., Chakraborty, S., and Ganguly, N. (2021) Unsupervised Topology Assessment in Smart Homes. In 8th ACM IKDD CODS and 26th COMAD, pp. 193–197, Association for Computing Machinery, Bangalore, India.Nowadays, a wide range of IOT devices are deployed in a variety of environments and settings to enhance the quality of human life. With a huge amount of data being generated from them, privacy is becoming a very big concern. To determine the level of privacy breach that can be achieved, we introduce in this paper, an unsupervised approach to visualize the sensor network, which in turn divulges the indoor topology of a smart home. The results are obtained from a smart environment by conducting a series of deductions and analysis on sensor datasets generated by a smart home. The experimental results demonstrate that our approach is able to deduce room-level sensor topology for a smart home even without the knowledge of any activity label or any prior information about the environment.
@inproceedings{10.1145/3430984.3431028,
abstract = {Nowadays, a wide range of IOT devices are deployed in a variety of environments and settings to enhance the quality of human life. With a huge amount of data being generated from them, privacy is becoming a very big concern. To determine the level of privacy breach that can be achieved, we introduce in this paper, an unsupervised approach to visualize the sensor network, which in turn divulges the indoor topology of a smart home. The results are obtained from a smart environment by conducting a series of deductions and analysis on sensor datasets generated by a smart home. The experimental results demonstrate that our approach is able to deduce room-level sensor topology for a smart home even without the knowledge of any activity label or any prior information about the environment.},
address = {New York, NY, USA},
author = {Mukherjee, Avirup and Mallick, Madhumita and Chakraborty, Sandip and Ganguly, Niloy},
booktitle = {8th ACM IKDD CODS and 26th COMAD},
keywords = {leibnizailab myown},
pages = {193–197},
publisher = {Association for Computing Machinery},
series = {CODS COMAD 2021},
title = {Unsupervised Topology Assessment in Smart Homes},
year = 2021
}%0 Conference Paper
%1 10.1145/3430984.3431028
%A Mukherjee, Avirup
%A Mallick, Madhumita
%A Chakraborty, Sandip
%A Ganguly, Niloy
%B 8th ACM IKDD CODS and 26th COMAD
%C New York, NY, USA
%D 2021
%I Association for Computing Machinery
%P 193–197
%R 10.1145/3430984.3431028
%T Unsupervised Topology Assessment in Smart Homes
%U https://doi.org/10.1145/3430984.3431028
%X Nowadays, a wide range of IOT devices are deployed in a variety of environments and settings to enhance the quality of human life. With a huge amount of data being generated from them, privacy is becoming a very big concern. To determine the level of privacy breach that can be achieved, we introduce in this paper, an unsupervised approach to visualize the sensor network, which in turn divulges the indoor topology of a smart home. The results are obtained from a smart environment by conducting a series of deductions and analysis on sensor datasets generated by a smart home. The experimental results demonstrate that our approach is able to deduce room-level sensor topology for a smart home even without the knowledge of any activity label or any prior information about the environment.
%@ 9781450388177 - 1.Nandy, A., Sharma, S., Maddhashiya, S., Sachdeva, K., Goyal, P., and Ganguly, N. (2021) Question Answering over Electronic Devices: A New Benchmark Dataset and a Multi-Task Learning based QA Framework, pp. 4600–4609, Association for Computational Linguistics.Answering questions asked from instructional corpora such as E-manuals, recipe books, etc., has been far less studied than open-domain factoid context-based question answering. This can be primarily attributed to the absence of standard benchmark datasets. In this paper, we meticulously create a large amount of data connected with E-manuals and develop a suitable algorithm to exploit it. We collect E-Manual Corpus, a huge corpus of 307,957 E-manuals, and pretrain RoBERTa on this large corpus. We create various benchmark QA datasets which include question answer pairs curated by experts based upon two E-manuals, real user questions from Community Question Answering Forum pertaining to E-manuals etc. We introduce EMQAP (E-Manual Question Answering Pipeline) that answers questions pertaining to electronics devices. Built upon the pretrained RoBERTa, it harbors a supervised multi-task learning framework which efficiently performs the dual tasks of identifying the section in the E-manual where the answer can be found and the exact answer span within that section. For E-Manual annotated question-answer pairs, we show an improvement of about 40% in ROUGE-L F1 scores over most competitive baseline. We perform a detailed ablation study and establish the versatility of EMQAP across different circumstances. The code and datasets are shared at https://github.com/abhi1nandy2/EMNLP-2021-Findings, and the corresponding project website is https://sites.google.com/view/emanualqa/home.
@proceedings{nandy2021question,
abstract = {Answering questions asked from instructional corpora such as E-manuals, recipe books, etc., has been far less studied than open-domain factoid context-based question answering. This can be primarily attributed to the absence of standard benchmark datasets. In this paper, we meticulously create a large amount of data connected with E-manuals and develop a suitable algorithm to exploit it. We collect E-Manual Corpus, a huge corpus of 307,957 E-manuals, and pretrain RoBERTa on this large corpus. We create various benchmark QA datasets which include question answer pairs curated by experts based upon two E-manuals, real user questions from Community Question Answering Forum pertaining to E-manuals etc. We introduce EMQAP (E-Manual Question Answering Pipeline) that answers questions pertaining to electronics devices. Built upon the pretrained RoBERTa, it harbors a supervised multi-task learning framework which efficiently performs the dual tasks of identifying the section in the E-manual where the answer can be found and the exact answer span within that section. For E-Manual annotated question-answer pairs, we show an improvement of about 40% in ROUGE-L F1 scores over most competitive baseline. We perform a detailed ablation study and establish the versatility of EMQAP across different circumstances. The code and datasets are shared at https://github.com/abhi1nandy2/EMNLP-2021-Findings, and the corresponding project website is https://sites.google.com/view/emanualqa/home.},
address = {Association for Computational Linguistics},
author = {Nandy, Abhilash and Sharma, Soumya and Maddhashiya, Shubham and Sachdeva, Kapil and Goyal, Pawan and Ganguly, NIloy},
howpublished = {Punta Cana, Dominican Republic},
keywords = {leibnizailab myown},
pages = {4600-4609},
title = {Question Answering over Electronic Devices: A New Benchmark Dataset and a Multi-Task Learning based QA Framework},
year = 2021
}%0 Conference Proceedings
%1 nandy2021question
%A Nandy, Abhilash
%A Sharma, Soumya
%A Maddhashiya, Shubham
%A Sachdeva, Kapil
%A Goyal, Pawan
%A Ganguly, NIloy
%C Association for Computational Linguistics
%D 2021
%P 4600-4609
%T Question Answering over Electronic Devices: A New Benchmark Dataset and a Multi-Task Learning based QA Framework
%U https://aclanthology.org/2021.findings-emnlp.392
%X Answering questions asked from instructional corpora such as E-manuals, recipe books, etc., has been far less studied than open-domain factoid context-based question answering. This can be primarily attributed to the absence of standard benchmark datasets. In this paper, we meticulously create a large amount of data connected with E-manuals and develop a suitable algorithm to exploit it. We collect E-Manual Corpus, a huge corpus of 307,957 E-manuals, and pretrain RoBERTa on this large corpus. We create various benchmark QA datasets which include question answer pairs curated by experts based upon two E-manuals, real user questions from Community Question Answering Forum pertaining to E-manuals etc. We introduce EMQAP (E-Manual Question Answering Pipeline) that answers questions pertaining to electronics devices. Built upon the pretrained RoBERTa, it harbors a supervised multi-task learning framework which efficiently performs the dual tasks of identifying the section in the E-manual where the answer can be found and the exact answer span within that section. For E-Manual annotated question-answer pairs, we show an improvement of about 40% in ROUGE-L F1 scores over most competitive baseline. We perform a detailed ablation study and establish the versatility of EMQAP across different circumstances. The code and datasets are shared at https://github.com/abhi1nandy2/EMNLP-2021-Findings, and the corresponding project website is https://sites.google.com/view/emanualqa/home. - 1.Adhisantoso, Y. G., Rohlfing, C., Voges, J., and Ostermann, J. (2020) Method for the coding of genotype likelihood of variant m55356, ISO/IEC JTC 1/SC 29/WG 8.
@article{AdhR2020b,
author = {Adhisantoso, Yeremia Gunawan and Rohlfing, Christian and Voges, Jan and Ostermann, Jörn},
journal = {ISO/IEC JTC 1/SC 29/WG 8},
keywords = {Method genotype leibnizailab likelihood myown of variant},
month = {oct},
title = {Method for the coding of genotype likelihood of variant m55356},
year = 2020
}%0 Journal Article
%1 AdhR2020b
%A Adhisantoso, Yeremia Gunawan
%A Rohlfing, Christian
%A Voges, Jan
%A Ostermann, Jörn
%D 2020
%J ISO/IEC JTC 1/SC 29/WG 8
%T Method for the coding of genotype likelihood of variant m55356 - 1.Adhisantoso, Y. G., Rohlfing, C., Voges, J., and Ostermann, J. (2020) Extension to method for the coding of genomic variants m55355, ISO/IEC JTC 1/SC 29/WG 8.
@article{AdhR2020,
author = {Adhisantoso, Yeremia Gunawan and Rohlfing, Christian and Voges, Jan and Ostermann, Jörn},
journal = {ISO/IEC JTC 1/SC 29/WG 8},
keywords = {Extension genomic leibnizailab myown variants},
month = {oct},
title = {Extension to method for the coding of genomic variants m55355},
year = 2020
}%0 Journal Article
%1 AdhR2020
%A Adhisantoso, Yeremia Gunawan
%A Rohlfing, Christian
%A Voges, Jan
%A Ostermann, Jörn
%D 2020
%J ISO/IEC JTC 1/SC 29/WG 8
%T Extension to method for the coding of genomic variants m55355 - 1.Minh, C. N. D., Gilani, S. Z., Islam, S., and Suter, D. (2020) Learning Affordance Segmentation: An Investigative Study. In DICTA2020.
@inproceedings{chauDICTA2020,
author = {Minh, Chau Nguyen Duc and Gilani, Syed Zulqarnain and Islam, Syed and Suter, David},
booktitle = {DICTA2020},
keywords = {leibnizailab myown},
title = {Learning Affordance Segmentation: An Investigative Study},
year = 2020
}%0 Conference Paper
%1 chauDICTA2020
%A Minh, Chau Nguyen Duc
%A Gilani, Syed Zulqarnain
%A Islam, Syed
%A Suter, David
%B DICTA2020
%D 2020
%R 10.1109/DICTA51227.2020.9363390
%T Learning Affordance Segmentation: An Investigative Study - 1.Tan, D., Maybery, M., Gilani, S. Z., Alvares, G., Mian, A., Suter, D., and Whitehouse, A. (2020) A broad autism phenotype expressed in facial morphology, Translational Psychiatry 10.
@article{TanPsych2019,
author = {Tan, Diana and Maybery, Murray and Gilani, Syed Zulqarnain and Alvares, Gail and Mian, Ajmal and Suter, David and Whitehouse, Andrew},
journal = {Translational Psychiatry},
keywords = {leibnizailab myown},
number = 1,
title = {A broad autism phenotype expressed in facial morphology},
volume = 10,
year = 2020
}%0 Journal Article
%1 TanPsych2019
%A Tan, Diana
%A Maybery, Murray
%A Gilani, Syed Zulqarnain
%A Alvares, Gail
%A Mian, Ajmal
%A Suter, David
%A Whitehouse, Andrew
%D 2020
%J Translational Psychiatry
%N 1
%R 10.1038/s41398-020-0695-z
%T A broad autism phenotype expressed in facial morphology
%V 10 - 1.Fayyazifar, N., Ahderom, S., Suter, D., Maiorana, A., and Dwivedi, G. (2020) Impact of Neural Architecture Design on Cardiac Abnormality Classification Using 12-lead ECG Signals. In 2020 Computing in Cardiology, pp. 1–4.
@inproceedings{9344384,
author = {Fayyazifar, N. and Ahderom, S. and Suter, D. and Maiorana, A. and Dwivedi, G.},
booktitle = {2020 Computing in Cardiology},
keywords = {leibnizailab myown},
pages = {1-4},
title = {Impact of Neural Architecture Design on Cardiac Abnormality Classification Using 12-lead ECG Signals},
year = 2020
}%0 Conference Paper
%1 9344384
%A Fayyazifar, N.
%A Ahderom, S.
%A Suter, D.
%A Maiorana, A.
%A Dwivedi, G.
%B 2020 Computing in Cardiology
%D 2020
%P 1-4
%R 10.22489/CinC.2020.161
%T Impact of Neural Architecture Design on Cardiac Abnormality Classification Using 12-lead ECG Signals