• Publications
  • Influence
AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty
TLDR
AugMix significantly improves robustness and uncertainty measures on challenging image classification benchmarks, closing the gap between previous methods and the best possible performance in some cases by more than half.
The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalization
TLDR
It is found that using larger models and artificial data augmentations can improve robustness on real-world distribution shifts, contrary to claims in prior work.
MNIST-C: A Robustness Benchmark for Computer Vision
TLDR
This work demonstrates that several previously published adversarial defenses significantly degrade robustness as measured by MNIST-C, a comprehensive suite of 15 corruptions applied to the MNIST test set, and hopes that this benchmark serves as a useful tool for future work in designing systems that are able to learn robust feature representations that capture the underlying semantics of the input.
SLIP: Self-supervision meets Language-Image Pre-training
TLDR
This work introduces SLIP, a multi-task learning framework for combining self-supervised learning and CLIP pre-training and finds that SLIP enjoys the best of both worlds: better performance than self- supervision and language supervision.
Parameter Re-Initialization through Cyclical Batch Size Schedules
TLDR
This work proposes a method of weight re-initialization by repeated annealing and injection of noise in the training process motivated by a Bayesian perspective of neural network training to improve language modeling performance and reduce training iterations.
Applying Text Analytics to the Mind-section Literature of the Tibetan Tradition of the Great Perfection
TLDR
This work uses the lens of text analytics tools based on machine learning techniques to investigate a number of questions of interest to scholars of this and related traditions of the Great Perfection.
Defending against Adversarial Patches with Robust Self-Attention
TLDR
A new defense against adversarial patch attacks based on the proposed Robust SelfAttention (RSA) layer, which replaces the outlier-sensitive weighted mean operation used by standard Self-Attention with a robust aggregation mechanism that detects and masks outlier tokens.
AUGMIX: A SIMPLE DATA PROCESSING METHOD
TLDR
AUGMIX significantly improves robustness and uncertainty measures on challenging image classification benchmarks, closing the gap between previous methods and the best possible performance in some cases by more than half.
A Critical Analysis of Distribution Shift
TLDR
It is found that some methods consistently help with distribution shifts in texture and local image statistics, but these methods do not help with some other distribution shifts like geographic changes, so no evaluated method consistently improves robustness.