Bio:

I am a Machine Learning Researcher, a native of Assam (অসম), India.  A few of my professional links:
Google Scholar LinkedIn dblp CSAuthorsNet GitHub Other Website (not actively maintained)

My ERDOS PATH reads as:
Ujjal Kr Dutta -> Mehrtash Harandi -> Richard Hartley -> Marc Kilgour -> Peter C. Fishburn -> Paul Erdos.

Research Interests:

List of my research publications can be found here.

Broad  Research  Agenda:

• Advance efficient and generalizable machine learning methods that require less human supervision, adapt incrementally, and transfer knowledge across domains.

• Focus area: Fundamental/basic and applied ML research (for solving real-world problems). 

Research Topics I have worked on:

• Representation & Metric Learning: Advanced methods for learning robust, transferable representations, including manifold learning, graph-based learning, adversarial learning, and Riemannian optimization.

• Domain Generalization & Transfer Learning: Developed domain-invariant models for vision tasks, enabling robust performance across unseen data distributions.

• Semi-/Unsupervised & Continual Learning: Pioneered techniques for learning from limited labels, and for incremental adaptation—crucial for lifelong AI agents.

• Applied Computer Vision and Machine Learning: Led projects on object detection, segmentation, pose estimation, 3D vision, virtual try-on, satellite imagery based Earth Observation (EO), time-series and sequential modeling for sensor data, recommender systems, multi-armed and contextual bandits.

• Generative Models & Diffusion: Explored GAN, VAE, Stable Diffusion, and neural fields for content generation and augmentation.

• Model Compression & Knowledge Distillation: Designed efficient, low-latency models for edge deployment.

• Physics-Informed ML: Integrated domain knowledge into neural models for scientific and industrial applications.

Professional Activities:

• Program Committee Member / Reviewer for

  • International Conferences:

    • International Conference on Learning Representations (ICLR) 2025

    • IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2022, 2023 

    • International Conference on Computer Vision (ICCV) 2023

    • European Conference on Computer Vision (ECCV) 2022, 2024

    • Neural Information Processing Systems (NeurIPS) 2024

    • AAAI Conference on Artificial Intelligence (AAAI) 2021 , 2022

    • Winter Conference on Applications of Computer Vision (WACV) 2021

    • Manifold Learning from Euclid to Riemann (Manlearn) workshop at International Conference on Pattern Recognition ( ICPR ) 2020/21 

    • AISTATS 2021 (invited), 2023, 2024

    •  ICANN 2019: Link 

  • International Journals:  ( publons profile )

    • IEEE Transactions on Neural Networks and Learning Systems (IEEE TNNLS)

    • IEEE Transactions on Artificial Intelligence (IEEE TAI)

    • IEEE Transactions on Circuits and Systems for Video Technology (IEEE TCSVT)

    • Image and Vision Computing

    • IEEE Transactions on Geoscience and Remote Sensing (IEEE TGRS)

Other Interests:

Apart from research, I am an athlete in the following sports: Olympic-style Weightlifting, Powerlifting and Mixed Martial Arts (MMA)/ Muay Thai. Till my undergrad, I also used to play soccer (mid-field), which I discontinued eventually to focus on strength and combat sports. During undergrad,  I competed in my first set of Weightlifting and Powerlifting competitions where I got bronze and silver medals.

During my stay in the PhD programme, I got a lot of opportunities to participate, and win, in many of the competitions involving these sports. In particular, I was 4 times Strongman (2015, 2017, 2018, 2019) in IITM Powerlifting (for lifting the highest weight per pound of body weight. I have lifted 190kg deadlift, as my personal best), 1 time silver medalist representing IITM at Chennai district level Olympic-style Weightlifting competition (2015), and 1 time silver medalist at the national level Inter-IIT Tournament in Olympic-style Weightlifting (2017).

Have a look here.