Foundational challenges in assuring alignment and safety of large language models
This work identifies 18 foundational challenges in assuring the alignment and safety of large
language models (LLMs). These challenges are organized into three different categories …
language models (LLMs). These challenges are organized into three different categories …
Provable guarantees for neural networks via gradient feature learning
Neural networks have achieved remarkable empirical performance, while the current
theoretical analysis is not adequate for understanding their success, eg, the Neural Tangent …
theoretical analysis is not adequate for understanding their success, eg, the Neural Tangent …
Learning a neuron by a shallow relu network: Dynamics and implicit bias for correlated inputs
We prove that, for the fundamental regression task of learning a single neuron, training a
one-hidden layer ReLU network of any width by gradient flow from a small initialisation …
one-hidden layer ReLU network of any width by gradient flow from a small initialisation …
Adversarial Robustness of In-Context Learning in Transformers for Linear Regression
Transformers have demonstrated remarkable in-context learning capabilities across various
domains, including statistical learning tasks. While previous work has shown that …
domains, including statistical learning tasks. While previous work has shown that …
Trained transformer classifiers generalize and exhibit benign overfitting in-context
Transformers have the capacity to act as supervised learning algorithms: by properly
encoding a set of labeled training (" in-context") examples and an unlabeled test example …
encoding a set of labeled training (" in-context") examples and an unlabeled test example …
Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data
Adversarial training is a widely-applied approach to training deep neural networks to be
robust against adversarial perturbation. However, although adversarial training has …
robust against adversarial perturbation. However, although adversarial training has …
Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networks
In this work, we investigate a particular implicit bias in the gradient descent training process,
which we term" Feature Averaging", and argue that it is one of the principal factors …
which we term" Feature Averaging", and argue that it is one of the principal factors …
Optimization dependent generalization bound for ReLU networks based on sensitivity in the tangent bundle
Recent advances in deep learning have given us some very promising results on the
generalization ability of deep neural networks, however literature still lacks a comprehensive …
generalization ability of deep neural networks, however literature still lacks a comprehensive …
Can Implicit Bias Imply Adversarial Robustness?
The implicit bias of gradient-based training algorithms has been considered mostly
beneficial as it leads to trained networks that often generalize well. However, Frei et …
beneficial as it leads to trained networks that often generalize well. However, Frei et …
MALT Powers Up Adversarial Attacks
Current adversarial attacks for multi-class classifiers choose the target class for a given input
naively, based on the classifier's confidence levels for various target classes. We present a …
naively, based on the classifier's confidence levels for various target classes. We present a …