I'm a PhD student in Statistics & Data Science at Yale.
My research studies the foundations of machine intelligence, with an emphasis on generalization, representation, and learning.
I explore these themes through complementary theoretical analysis and empirical investigation:
- Deep learning, representation learning, & inductive structure: Developing novel methods and architectures to improve systematic compositional generalization and data efficiency, sometimes drawing inspiration from biological intelligence to achieve human-like reasoning and out-of-distribution generalization.
- Theory of modern learning systems: Developing frameworks that explain empirical phenomena in contemporary machine learning through unified statistical and computational principles, aiming to develop a foundation for future progress in artificial intelligence.
Where to start: If you're interested in neural network architectures, check out our work on an extension of the transformer architecture with explicit relational mechanisms and inductive biases (blog ⧉). For theoretical analysis of modern machine learning methods, see our statistical learning theory framework for chain-of-thought supervised learning (blog ⧉).
Pinned Loading
-
abstractor
abstractor PublicThis is the code repository associated with the paper "Abstractors and relational cross-attention: An inductive bias for explicit relational reasoning in Transformers"
-
hierarchical-attention-external-memory
hierarchical-attention-external-memory PublicHierarchical attention mechanisms for external memory in machine learning models.
Jupyter Notebook
-
high-dim-universality-erm
high-dim-universality-erm PublicNumerical experiments exploring the "universality" of empirical risk minimization in high dimensions.
HTML
-
relational-convolutions
relational-convolutions PublicThis is the project repo associated with the paper "Relational Convolutional Networks: A framework for learning representations of hierarchical relations" by Awni Altabaa, John Lafferty
-
abstract_transformer
abstract_transformer PublicThis is the project repo associated with the paper "Disentangling and Integrating Relational and Sensory Information in Transformer Architectures" by Awni Altabaa, John Lafferty
Jupyter Notebook 6
If the problem persists, check the GitHub status page or contact support.