Aamer Abdul

I am a Masters student at Mila Quebec where I am advised by Professor Samira Ebrahimi Kahou. Currently, my research interests lie at the intersection of generalization, explainabiltiy and robustness, applied towards problems in climate and healthcare. I am also interested offline reinforcement learning and vision-language models.

Previously, I was a ML research intern at the University of Toronto, working under the supervision of Professor Chi-Guhn Lee, where I worked on adversarial domain adaptation for data-sparse object detection. My undergrad was in Electronics and Communication Engineering from BITS Pilani, during which I mainly worked on computer vision and machine listening.

I'm actively seeking internship opportunities and am open to collaborations. Don't hesitate to get in touch!

Email  /  Resume  /  Publications  /  Google Scholar  /  Twitter  /  LinkedIn


News

Publications
Empowering Clinicians with MeDT: A Framework for Sepsis Treatment
Aamer Abdul Rahman, Pranav Agarwal, Rita Noumeir, Philippe Jouvet, Vincent Michalski, Samira Ebrahimi Kahou
Submitted to TMLR 2024
NeurIPS 2023 Goal-Conditioned Reinforcement Learning Workshop (Spotlight)

Project page / Paper / Code

Offline reinforcement learning is promising for safety-critical tasks like clinical decision support, but faces challenges of interpretability and clinician interactivity. To overcome these, the proposed Medical Decision Transformer (MeDT) utilizes a goal-conditioned RL paradigm for sepsis treatment recommendations. MeDT employs the decision transformer architecture, considering factors like treatment outcomes, patient acuity scores, dosages, and current/past medical states to provide a holistic view of the patient's history. This enhances decision-making by allowing MeDT to generate actions based on user-specified goals, ensuring clinician interactability and addressing sparse rewards. Results from the MIMIC-III dataset demonstrate MeDT's effectiveness in producing interventions that either outperform or compete with existing methods, offering a more interpretable, personalized, and clinician-directed approach.

Transformers in Reinforcement Learning: A Survey
Pranav Agarwal, Aamer Abdul Rahman, Pierre-Luc St-Charles, Simon J.D. Prince, Samira Ebrahimi Kahou
Arxiv Preprint, 2023

Paper

We present a comprehensive survey on the emergence of transformers in reinforcement learning.

Revisiting Learnable Affines for Batch Norm in Few-Shot Transfer Learning
Moslem Yazdanpanah*, Aamer Abdul Rahman*, Muawiz Chaudhary, Christian Desrosiers, Mohammad Havaei, Eugene Belilovsky, Samira Ebrahimi Kahou
CVPR, 2022

Paper / Code

We introduce a novel normalization strategy for few-shot transfer learning, bringing performance improvements upto 5% in accuracy with neglible computational overhead.

Pitfalls of Conditional Batch Normalization for Contextual Multi-Modal Learning
Ivaxi Sheth, Aamer Abdul Rahman, Mohammad Havaei, Samira Ebrahimi Kahou
NeurIPS, 2022, ICBINB Workshop

arxiv

We demonstrate that using conditional batch normalization for contextual multimodal tasks leads to shortcut learning. Our experiments reveal that the network learns close to no visual features and relies solely on contextual auxiliary data to make predictions.

Learning from Uncertain Concepts via Test Time Interventions
Ivaxi Sheth, Aamer Abdul Rahman, Laya Rafiee, Mohammad Havaei, Samira Ebrahimi Kahou
NeurIPS, 2022, Trustworthy and Socially Responsible Machine Learning Workshop

arxiv

We propose a novel uncertainty based strategy to select the interventions in Concept Bottleneck Models. We demonstrate the proposed method can improve human-model interaction via more meaningful corrections, while reducing information leakage.

Robust Segmentation of Vascular Network using Deeply Cascaded AReN-UNet
Aamer Abdul Rahman, Birendra Biswal, Geetha Pavani, Shazia Hasan
Biomedical Signal Processing and Control, 2021

We present a U-Net segmentation architecture based on cascading and attention modules for improved performance on retinal blood vessel segmentation.

Classification of Urbansound8k: A Study using Convolutional Neural Networks and Multiple Data Augmentation Techniques
Aamer Abdul Rahman, Angel Arul Jyothi
International Conference on Soft Computing and its Engineering Applications, 2020

A study on effectiveness of data augmentation strategies on downstream performance for urban sound classification with CNNs.

Early Detection of Locust Swarms using Deep Learning
Karthika Suresh, Aamer Abdul Rahman
Advances in Machine Learning and Computational Intelligence, 2020

An object detection approach with Faster RCNN's to detect and track locust insects.


The template is borrowed from Jon Barron's homepage