A key factor in the success of deep neural networks is the ability to scale models to improve
performance by varying the architecture depth and width. This simple property of neural …

Existing work on understanding deep learning often employs measures that compress all
data-dependent information into a few numbers. In this work, we adopt a perspective based …

Machine learning models are vulnerable to membership inference attacks in which an
adversary aims to predict whether or not a particular sample was contained in the target …

Investigating active learning, we focus on the relation between the number of labeled
examples (budget size), and suitable querying strategies. Our theoretical analysis shows a …

Current deep learning methods are regarded as favorable if they empirically perform well on
dedicated test sets. This mentality is seamlessly reflected in the resurfacing area of continual …

Understanding AI systems' inner workings is critical for ensuring value alignment and safety.
This review explores mechanistic interpretability: reverse-engineering the computational …

The Lean Data Scientist: Recent Advances Toward Overcoming the Data Bottleneck Page 1
OBTAINING DATA HAS become the key bottleneck in many machine-learning (ML) applications …

Researchers investigating example hardness have increasingly focused on the dynamics by
which neural networks learn and forget examples throughout training. Popular metrics …

Pretrained models are the standard starting point for training. This approach consistently
outperforms the use of a random initialization. However, pretraining is a costly endeavour …

Post-hoc gradient-based interpretability methods [Simonyan et al., 2013, Smilkov et al.,
2017] that provide instance-specific explanations of model predictions are often based on …