Neil Band

I'm a third-year PhD student in the Computer Science Department at Stanford University. My advisors are Tatsunori Hashimoto and Tengyu Ma. My work is supported by the NSF GRFP and Quad Fellowship.

I was previously a Masters by Research student in the Oxford Applied and Theoretical Machine Learning Group (OATML) at the University of Oxford, where I was advised by Yarin Gal and supported by a Rhodes Scholarship. Before that, I did my A.B. in Computer Science and Economics at Harvard College. I worked with Stratos Idreos in the Harvard Data Systems Lab on systems machine learning, and Manolis Kellis in the MIT Computational Biology Group on machine learning in single-cell transcriptomics.

Google Scholar  /  Twitter  /  Github

My long-term goal is to build frontier AI systems that safely and legibly advance scientific knowledge and our collective well-being. My current focus is on improving the mathematical and general-domain reasoning capabilities of large language models.

A simple approach to learn knowledge from corpora as small as ~1M tokens, using graph-based synthetic data generation followed by continued pretraining.

An alignment objective to train language models that express calibrated verbal statements of confidence in their long-form natural language generations.

An easy-to-use, expert-guided, open-source suite of diabetic retinopathy detection benchmarking tasks for Bayesian deep learning.

Three industry-scale tasks for evaluating robustness and uncertainty quality under distribution shift, including the largest vehicle motion prediction dataset to date.

A novel deep learning architecture that takes the entire dataset as input and learns to reason about relationships between datapoints using self-attention.

A set of tasks assessing the reliability of models on both vision and language, and Plex, a large pretrained model which improves SoTA across tasks.

A high-quality repository of many datasets, baselines, and techniques relevant to the Bayesian deep learning literature.

An optimization framework for distributed deep learning that jointly considers memory usage and computation time when searching for a parallelization strategy.

Based on Jon Barron's website (source code here).