AI Research Scientist, Meta AI Dec '22 - Present
Research on large language models in the Llama team
I am a Research Scientist in the LLama Research team at Meta. Previously, I was a Principal Research Scientist and a Distinguished Member of Technical Staff (DMTS) at Bell Labs.
At Meta, I work on large language models, leading projects aimed at improving the reasoning capabilities of Llama models. In addition, I am also interested in the topics of Self-Supervised Learning, Federated Learning, On-Device ML, and domain adaptation. My research has been covered by several media organizations including the New Yorker, Financial Times, Livemint and Canadian Broadcasting Corporation. I also serve on the Editorial Board of ACM IMWUT journal as an Associate Editor.
I hold a Ph.D. in Machine Learning from the University College London while also being affiliated with the Cambridge Machine Learning Systems Lab (CaMLSys). My PhD advisors were Prof. Nic Lane and Prof. Nadia Berthouze. I also hold a Masters in Computer Science from the University of Toronto and a B.Tech. from DA-IICT where I was awarded the President's Gold Medal.
[P62] Llama Team, AI @ Meta. "The Llama 3 Herd of Models".
[P61] Deldari S., Spathis D., Malekzadeh M., Kawsar F., Salim F.D., Mathur A.. "Crossl: Cross-modal self-supervised learning for time-series through latent masking". ACM WSDM, 2024
[P60] Tang C.I., Qendro L., Spathis D., Kawsar F., Mascolo C., Mathur A.. "Kaizen: Practical self-supervised continual learning with continual fine-tuning". IEEE/CVF Winter Conference on Applications of Computer Vision
[P59] Qiu X., Parcollet T., Fernandez-Marques J., Gusmão P.P.B, Gao Y., Beutel D., Topal T., Mathur A., Lane N.D. "A First Look into the Carbon Footprint of Federated Learning". Journal of Machine Learning Research 24 (2023) 1-23
[P58] Hutiri W., Ding A., Kawsar F., Mathur A. "Tiny, Always-on and Fragile: Bias Propagation through Design Choices in On-device Machine Learning Workflows". ACM Transactions on Software Engineering and Methodology 2023.
[P57] Lubana E.S., Tang C.I., Kawsar F., Dick R., Mathur A.. "Orchestra: Unsupervised Federated Learning via Globally Consistent Clustering". ICML 2022 (Spotlight Talk)
[P56] Cho H., Mathur A., Kawsar F. "FLAME: Federated Learning Across Multi-device Environments". ACM IMWUT, 2022.
[P55] Jain Y., Tang C.I., Min C., Kawsar F., and Mathur A. "ColloSSL: Collaborative Self-Supervised Learning for Human Activity Recognition". ACM IMWUT, 2022.
[P54] Gan S., Mathur A., Isopoussu A., Kawsar F., Berthouze N., Lane N.D. "FRuDA: Framework for Distributed Adversarial Domain Adaptation". IEEE Transactions on Parallel And Distributed Systems, 2022.
[P53] Toussaint W., Mathur A., Ding A., Kawsar F. "Characterising the Role of Pre-Processing Parameters in Audio-based Embedded Machine Learning". AIChallengeIoT, ACM SenSys 2021.
[P52] Cho H., Mathur A., Kawsar F. "Device or User: Rethinking Federated Learning in Personal-Scale Multi-Device Environments". AIChallengeIoT, ACM SenSys 2021.
[P51] Franklin M., Lagnado D., Min C., Mathur A., Kawsar F. "Designing Memory Aids for Dementia Patients using Earables". EarComp 2021.
[P50] Wang C., Olugbade T., Mathur A., Williams A., Lane N.D., Berthouze N. "Chronic Pain Protective Behavior Detection with Deep Learning". ACM Transactions on Computing for Healthcare 2021.
[P49] Wang C., Gao Y., Mathur A., Williams A., Lane N.D., Berthouze N. "Leveraging activity recognition to enable protective behavior detection in continuous data". ACM IMWUT 2021.
[P48] Lei J., Rahman T., Shafik R., Wheeldon A., Yakovlev A., Granmo O., Kawsar F., and Mathur A.. "Low-Power Audio Keyword Spotting using Tsetlin Machines". Journal of Low Power Electronics and Applications.
[P47] Mathur A., Beutel, D.J., Gusmão P.P.B, Fernandez-Marques J., Topal, T., Qiu, X., Parcollet, T., Gao Y., and Lane, N.D. "On-device Federated Learning with Flower". On-Device Intelligence Workshop at MLSys 2021.
[P46] Qiu X., Parcollet T., Beutel D., Topal T., Mathur A., Lane N.D. "Can Federated Learning Save the Planet?"" Tackling Climate Change with Machine Learning workshop at NeurIPS 2020
[P45] Jain Y., Tang C.I., Min C., Kawsar F., and Mathur A. "Group Supervised Learning: Extending Self-Supervised Learning to Multi-Device Settings". ICML 2021 workshop on Self-Supervised Learning for Reasoning and Perception.
[P44] Mathur A., Berthouze N., Lane N.D. "Unsupervised Domain Adaptation under Label Space Mismatch for Speech Classification". Interspeech 2020.
[P43] Beutel, D.J., Topal, T., Mathur A., Qiu, X., Parcollet, T. and Lane, N.D. "Flower: A Friendly Federated Learning Research Framework". arXiv preprint arXiv:2007.14390.
[P42] Mathur A., Kawsar F., Berthouze N., Lane N.D. "LIBRI-ADAPT: A new speech dataset for unsupervised domain adaptation". IEEE ICASSP 2020.
[P41] Chang Y., Mathur A., Isopoussu A., Song J., Kawsar F. "A Systematic Study of Unsupervised Domain Adaptation for Robust Human-Activity Recognition". ACM IMWUT 2020.
[P40] Mathur A., Isopoussu A., Kawsar F., Berthouze N., Lane N.D. "FlexAdapt: Flexible Cycle-Consistent Adversarial Domain Adaptation". IEEE International Conference on Machine Learning and Applications, 2019.
[P39] Gong T., Ramos A.G.C.P, Bhattacharya S., Mathur A., Kawsar F. "AudioDOS: Real-Time Denial-of-Service Adversarial Attacks on Deep Audio Models". IEEE International Conference on Machine Learning and Applications, 2019.
[P38] Min C., Montanari A., Mathur A., and Kawsar F. "A Closer Look at Quality-Aware Runtime Assessment of Sensing Models in Multi-Device Environments", ACM Sensys 2019.
[P37] Antonini M., Vu T., Min C., Montanari A., Mathur A., and Kawsar F. "Resource Characterisation of Personal-Scale Sensing Models on Edge Accelerators", AIChallengeIoT, ACM SenSys 2019.
[P36] Mathur A., Isopoussu A., Berthouze N., Lane N.D., Kawsar F. "Unsupervised Domain Adaptation for Robust Sensory Systems". CML-IoT, Ubicomp 2019, London, UK.
[P35] Min C., Mathur A., Montanari A., Kawsar F. "An Early Characterisation of Wearing Variability on Motion Signals for Wearables". ISCW 2019, London, UK.
[P34] Wang C., Olugbade T., Mathur A., Williams A., Berthouze N., Lane N.D. "Recurrent Network based Automatic Detection of Chronic Pain Protective Behavior using MoCap and sEMG data". ISCW 2019, London, UK.
[P33] Katayama S., Mathur A., van den Broeck M., Okoshi T., Nakazawa J., Kawsar F. "Situation-Aware Emotion Regulation of Conversational Agents with Kinetic Earables". ACII 2019, Cambridge, UK.
[P32] Mathur A., Isopoussu A., Kawsar F., Berthouze N., Lane N.D. "Mic2Mic: Using Cycle-Consistent Generative Adversarial Networks to Overcome Microphone Variability in Speech Systems". IPSN 2019, Montreal, Canada.
[P31] Mathur A., Kawsar F., Berthouze N., Lane N.D. "A Vision for Adaptive and Generalizable Audio-Sensing Systems". To Appear in SocialSense @ IPSN 2019.
[P30] Lee S., Min C., Montanari A., Mathur A. , Chang Y., Song J. and Kawsar F. "Automatic Smile and Frown Recognition with Kinetic Earables". The 10th Augmented Human International Conference (AH 2019), Reims Champagne-Ardenne, France.
[P29] Kawsar F., Min C., Mathur A. , Montanari A. "Earables for Personal-scale Behaviour Analytics", IEEE Pervasive Computing, Volume: 17, Issue: 3, 2018.
[P28] Mathur A., Isopoussu A., Kawsar F., Smith R., Lane N.D. and Berthouze N. "On Robustness of Cloud Speech APIs: An Early Characterization". In HASCA Workshop @ Ubicomp 2018.
[P27] Min C., Montanari A., Mathur A., Lee S., and Kawsar F. "Cross-Modal Approach for Conversational Well-being Monitoring with Multi-Sensory Earables", In Computing for WellBeing Workshop @ Ubicomp 2018.
[P26] Min C., Mathur A., Kawsar F. "Exploring Audio and Kinetic Sensing on Earable Devices", In WearSys Workshop @ Mobisys 2018.
[P25] Min C., Mathur A., Kawsar F. "Audio-Kinetic Model for Automatic Dietary Monitoring with Earable Devices", Mobisys 2018.
[P24] Mathur A., Zhang, T., Bhattacharya, S., Veličković, P., Joffe, L., Lane, N. D., Kawsar F, Lió, P. "A Deep Data Augmentation Training Method to Address Software and Hardware Heterogeneities in Wearable and Smartphone Sensing Devices”. IPSN 2018, Porto, Portugal.
[P23] Mathur A., Lane, N. D., Bhattacharya, S., Boran, A., Forlivesi, C., Kawsar, F. "DeepEye: Resource efficient local execution of multiple deep vision models using wearable commodity hardware". ACM Mobisys 2017.
[P22] Lane, N. D., Bhattacharya, S., Mathur A., Georgiev, P., Forlivesi, C., Kawsar, F. "Squeezing Deep Learning into Mobile and Embedded Devices". IEEE Pervasive Magazine 2017.
[P21] Mathur A., Kalanadhabhatta L.M., Majethia R., Kawsar F. "Moving Beyond Market Research: Demystifying Smartphone User Behavior in India". ACM IMWUT/Ubicomp 2017.
[P20] Lane, N., Bhattacharya, S., Mathur A., Forlivesi, C., Kawsar, F. "Dxtk: Enabling resource-efficient deep learning on mobile and embedded devices with the DeepX toolkit". EAI MobiCase 2016.
[P19] Mashhadi, A., Mathur A., Van den Broeck, M., Vanderhulst, G., Kawsar, F. "Understanding the impact of personal feedback on face-to-face interactions in the workplace". ACM ICMI 2016. (Best Paper Nominee)
[P18] Mathur A., Lane N.D., Kawsar F. "Engagement-aware computing: Modelling user engagement from mobile contexts”. ACM Ubicomp 2016.
[P17] Mathur A., Kawsar F. "The Need to Account for Geographical Diversities in Mobile Data Research”. MobiData @ MobiSys 2016.
[P16] Montanari A., Mashhadi A., Mathur A., Kawsar F. "Understanding the Privacy Design Space for Personal Connected Objects”. ACM BritishHCI 2016.
[P15] Mathur A., Broeck M., Vanderhulst G., Mashhadi A., and Kawsar F. "Tiny Habits in the Giant Enterprise: Understanding the Dynamics of a Quantified Workplace". ACM UbiComp 2015.
[P14] Mathur A., Broeck M., Vanderhulst G., Mashhadi A., and Kawsar F. "Quantified Workplace: Opportunities and Challenges". Physical Analytics Workshop @ MobiSys 2015.
[P13] Mashhadi A., Mathur A., Kawsar F. The Myth of Subtle Notifications. ACM Ubicomp 2014.
[P12] Mathur A., Jaiswal S. Exploring the Interplay between Community Media and Mobile Web in Developing Regions. ACM MobileHCI 2013. (Best Paper Honorable Mention Award)
[P11] Mathur A., Agarwal S., Jaiswal S. Exploring Playback and Recording of Web-based Audio Media on Low-End Feature Phones. ACM DEV 2012.
[P10] Samdaria N., Mathur A., Balakrishnan R. Paying in Kind for Crowdsourcing Work in Developing Regions. Pervasive 2012 (Best Paper Nominee)
[P9] Kumar N., Mathur A., Lal S. Banking 101: Mobile-izing financial inclusion in an emerging India. Bell Labs Technical Journal, Human Sciences and User Experience Edition, 2012
[P8] Mathur A., Majumder A., Datta S., Menon S. LifeView: A Lifelog Visualization Tool for Supporting Sentimental Recall and Sharing. ACM OzCHI 2012
[P7] Mathur A., Ramachandran D., Cutrell E., Balakrishnan R. An Exploratory Study on the Use of Cameraphones and Pico Projectors in Rural India. ACM MobileHCI 2011
[P6] Hutchful D., Mathur A., Joshi A., Cutrell A. Cloze: An Authoring Tool for Teachers with Low Computer Proficiency. IEEE/ACM ICTD '10
[P5] Kam M., Mathur A., Kumar A., Canny J. Designing Digital Games for Rural Children: A Study of Traditional Village Games in India. ACM CHI '09 (Best Paper Honorable Mention Award)
[P4] Kam M., Kumar A., Jain S., Mathur A., Canny J. Improving Literacy in Rural India: Cellphone Games in an After-School Program. IEEE/ACM ICTD '09
[P3] Kam M., Agarwal A., Kumar A., Lal S., Mathur A., Tewari A., Canny J. Designing E-Learning Games for Rural Children in India: A Format for Balancing Learning with Fun. ACM DIS '08
[P2] Kam M., Bhagwani S., Kumar A., Lal S., Mathur A., Tewari A., Canny J. The Social Complexities of User-Centered Design in ICTD: Experiences from Four Schools in India's Villages and Slums. IEEE/ACM ICTD '07
[P1] Tewari A., Kumar A., Mathur A., Lal S., Kam M., Canny J. Mobile Games for Learning English in Rural India: Designing Cellphone Games Informed by Traditional Games. 3rd Digital Games Research Association International Conference (DIGRA'07)
I have been fortunate to work closely with the following interns and student collaborators:
Research on large language models in the Llama team
I worked in the Pervasive Systems department where I led research on various machine learning topics, such as Federated Learning, Self-Supervised Learning, Algorithmic Fairness, and Domain Adaptation.