Bio
Nasik Muhammad Nafi is a Postdoctoral Research Associate in the Advanced Computing for Life Sciences & Engineering group at the National Center for Computational Sciences division. Prior to joining ORNL, Nafi obtained his Ph.D. in Computer Science from Kansas State University, Manhattan, KS in 2024. Previously, he received his Masters degree in Computer Science from Kansas State University and Bachelors degree in Computer Science and Engineering from Bangladesh University of Engineering and Technology. During his graduate studies, he worked as an intern at DEKA Research and Development as a Machine Learning Intern and at C2FO as a Data Scientist Intern.
Nafi's research focuses on improving generalization capability of machine learning models while learning from limited data, unlocking their potential to operate in complex, diverse, and dynamic real-world environments. His research interest broadly lies in the intersection of computer vision and deep reinforcement learning. His Ph.D. dissertation focused on developing algorithmic and architectural strategies to improve generalization in reinforcement learning. In particular, his research introduced efficient mechanisms that enables RL agent to transfer it's learning to take optimal decisions in previously unseen variations of the tasks. He also worked on analyzing the capability of generative models and developing spatio-temporal feature extraction framework for video processing. His research works are applied to plant disease identification, bee species classification, and risky tackle detection. At ORNL, he is currently working on developing transferable model for medical image segmentation and generalizable model for turbulence modeling problems. He was the best paper award winner at IEEE IWSSIP 2020. He has published conference and workshop papers in multiple venues such as NeurIPS, AAAI, AAMAS, IJCNN, ICMLA.
Awards
- Outstanding Graduate Student of the Month, College of Engineering, K-State, 2023.
- DeepMind Travel Grant, ICML 2022.
- Best Paper Award, IEEE IWSSIP 2020.
Education
- Ph.D. in Computer Science, Kansas State University, 2024.
- MS in Computer Science, Kansas State University, 2019.
- BS in Computer Science and Engineering, Kansas State University, 2015.
Professional Service
- Session Chair: IJCNN 2023, 2024.
- Co-chair: 4th AI - Diversity, Belonging, Equity, and Inclusion (AIDBEI) workshop at AAMAS 2022.
- Reviewer: NeurIPS 2024, AutoML 2022, NeurIPS Meta-Learning 2021, NeurIPS, Deep RL 2022, Artificial Intelligence Review Journal.
Specialized Equipment
- Programming: Python, C, C++, R.
- Machine Learning Frameworks: PyTorch, TensorFlow, Keras, scikit-learn.
- Reinforcement Learning Libraries: Garage, RLlib, OpenAI Baselines, Stable Baselines.
- RL/CV/NLP Tools: OpenAI Gym, Gymnasium, Mujoco, OpenCV, NLTK.
- Tools: VSCode, Jupyter Notebook, Google Colab, Docker.
Other Publications
- Nafi, N.M., Ali, R.F., Hsu, W., Duong, K. and Vick, M. (2024). "Policy Optimization with Horizon Regularized Advantage to Improve Generalization in Reinforcement Learning." in 23rd International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS).
- Nafi, N.M., Ali, R.F. and Hsu, W. (2024). "Analyzing the Sensitivity to Policy-Value Decoupling in Deep Reinforcement Learning Generalization." in 2024 International Joint Conference on Neural Networks (IJCNN).
- Nafi, N.M., Poggi-Corradini, G. and Hsu, W. (2023). "Policy Optimization with Augmented Value Targets for Generalization in Reinforcement Learning." in 2023 International Joint Conference on Neural Networks (IJCNN).
- Nafi, N.M., Glasscock, C. and Hsu, W. (2022). "Attention-based Partial Decoupling of Policy and Value for Generalization in Reinforcement Learning." in IEEE 21st International Conference on Machine Learning and Applications (ICMLA).
- Nafi, N.M., Dietrich, S. and Hsu, W. (2022). "Risky Tackle Detection from American Football Practice Videos using 3D Convolutional Networks." in 18th International Conference on Machine Learning and Data Mining (MLDM).
- Nafi, N.M. and Hsu, W.H. (2020). "Addressing Class Imbalance in Image-Based Plant Disease Detection: Deep Generative vs. Sampling-Based Approaches." in 27th International Conference on Systems, Signals and Image Processing (IWSSIP).
- Nafi, N.M., Bose, A., Khanal, S., Caragea, D. and Hsu, W.H. (2020). "Abstractive Text Summarization of Disaster-Related Documents." in 17th International Conference on Information Systems for Crisis Response and Management (ISCRAM).