📜 = Patent 📝 = Paper 💻 = Code 🔗 = Link
Mechanistic Interpretability
Mechanistic interpretability aims to reverse-engineer how machine learning models work by analyzing their internal computations. My research focuses on developing principled methods for attributing the behavior of thinking models, e.g. understanding how individual reasoning steps influence downstream computations and final outputs.
💻 Public repository for principled attribution research🔗 Interactive interface for causal attribution of multi-step reasoning in thinking modelsAI4Health
Applications of artificial intelligence in healthcare and medical diagnostics. For example: smartphone-based applications for automated interpretation of rapid diagnostic tests for HIV, syphilis, and COVID-19.
📜 Adaptable Automated Interpretation of Rapid Diagnostic Tests Using Self-Supervised Learning and Few-Shot Learning📝 SMARTtest: A Smartphone App to Facilitate HIV and Syphilis Self- and Partner-Testing, Interpretation of Results, and Linkage to Care📝 Rapidly Adaptable Automated Interpretation of Point-of-Care COVID-19 Diagnostics💻 Public repository for the machine learning pipelines of SMARTtest🔗 Example production machine learning pipeline for the INSTI test kitSTEM Education
Intelligent tutoring systems and informal learning methods to teach artificial intelligence to K-12.
📝 LogicLearner: A Tool for the Guided Practice of Propositional Logic Proofs📝 Hierarchical Multi-Armed Bandits for the Concurrent Intelligent Tutoring of Concepts and Problems of Varying Difficulty Levels📝 Teenagers and Artificial Intelligence: Bootcamp Experience and Lessons LearnedMedical Imaging with Deep Learning
Deep learning techniques for medical imaging. For example: longitudinal multiple sclerosis (MS) lesion segmentation and contrast-agnostic spinal cord segmentation.
📝 Contrast-agnostic Segmentation of the Spinal Cord Using Deep Learning📝 Team Neuropoly: Description of the Pipelines for the MICCAI 2021 MS New Lesions Segmentation Challenge📝 A Soft Segmentation Approach for New Multiple Sclerosis Lesion Detection (Poster)💻 Public repository for our submission to the MICCAI 2021 MS New Lesions Segmentation ChallengeComputational Neuroscience and Brain-Machine Interfaces
Formulation of computational models for understanding the brain and the development of brain-machine interfaces. For example: autonomous optimization of neuroprosthetic stimulation parameters for motor cortex and spinal cord outputs.
📝 Autonomous Optimization of Neuroprosthetic Stimulation Parameters That Drive the Motor Cortex and Spinal Cord Outputs in Rats and MonkeysRepresentation Learning
Investigating representation learning techniques for various application domains. For example: natural language processing and computer vision.
💻 Implementations for a family of attention mechanisms for applications in NLP💻 Comparatively finetuning BERT for various downstream NLP tasks💻 Implementations of several self-supervised pretext tasks for language and vision modalities💻 Simple, straight-forward extraction of acoustic and prosodic features from sound waves