research
∙
07/27/2023
Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback
Reinforcement learning from human feedback (RLHF) is a technique for tra...
share
research
∙
07/27/2023
Thinker: Learning to Plan and Act
We propose the Thinker algorithm, a novel approach that enables reinforc...
research
∙
04/19/2023
Investigating the Nature of 3D Generalization in Deep Neural Networks
Visual object recognition systems need to generalize from a set of 2D tr...
research
∙
03/16/2023
Characterizing Manipulation from AI Systems
Manipulation is a common concern in many domains, such as social media, ...
research
∙
03/10/2023
Unifying Grokking and Double Descent
A principled understanding of generalization in deep learning may requir...
research
∙
02/20/2023
Harms from Increasingly Agentic Algorithmic Systems
Research in Fairness, Accountability, Transparency, and Ethics (FATE) ha...
research
∙
02/03/2023
Blockwise Self-Supervised Learning at Scale
Current state-of-the-art deep networks are all powered by backpropagatio...
research
∙
01/09/2023
On The Fragility of Learned Reward Functions
Reward functions are notoriously difficult to specify, especially for ta...
research
∙
11/27/2022
Domain Generalization for Robust Model-Based Offline Reinforcement Learning
Existing offline reinforcement learning (RL) algorithms typically assume...
research
∙
11/15/2022
Mechanistic Mode Connectivity
Neural networks are known to be biased towards learning mechanisms that ...
research
∙
10/26/2022
Broken Neural Scaling Laws
We present a smoothly broken power law functional form that accurately m...
research
∙
10/06/2022
Towards Out-of-Distribution Adversarial Robustness
Adversarial robustness continues to be a major challenge for deep learni...
research
∙
09/27/2022
Defining and Characterizing Reward Hacking
We provide the first formal definition of reward hacking, a phenomenon w...
research
∙
09/20/2022
Metadata Archaeology: Unearthing Data Subsets by Leveraging Training Dynamics
Modern machine learning research relies on relatively few carefully cura...
research
∙
12/27/2021
Multi-Domain Balanced Sampling Improves Out-of-Distribution Generalization of Chest X-ray Pathology Prediction Models
Learning models that generalize under different distribution shifts in m...
research
∙
12/14/2021
Filling gaps in trustworthy development of AI
The range of application of artificial intelligence (AI) is vast, as is ...
research
∙
11/13/2020
Active Reinforcement Learning: Observing Rewards at a Cost
Active reinforcement learning (ARL) is a variant on reinforcement learni...
research
∙
09/19/2020
Hidden Incentives for Auto-Induced Distributional Shift
Decisions made by machine learning systems have increasing influence on ...
research
∙
05/30/2020
AI Research Considerations for Human Existential Safety (ARCHES)
Framed in positive terms, this report examines how technical AI research...
research
∙
04/15/2020
Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims
With the recent wave of progress in artificial intelligence (AI) has com...
research
∙
03/02/2020
Out-of-Distribution Generalization via Risk Extrapolation (REx)
Generalizing outside of the training distribution is an open challenge f...
research
∙
11/19/2018
Scalable agent alignment via reward modeling: a research direction
One obstacle to applying reinforcement learning algorithms to real-world...
research
∙
06/20/2018
Uncertainty in Multitask Transfer Learning
Using variational Bayes neural networks, we develop an algorithm capable...
research
∙
04/03/2018
Neural Autoregressive Flows
Normalizing flows and autoregressive models have been successfully combi...
research
∙
01/31/2018
Nested LSTMs
We propose Nested LSTMs (NLSTM), a novel RNN architecture with multiple ...
research
∙
12/13/2017
Deep Prior
The recent literature on deep learning offers new tools to learn a rich ...
research
∙
10/13/2017
Bayesian Hypernetworks
We propose Bayesian hypernetworks: a framework for approximate Bayesian ...
research
∙
06/16/2017
A Closer Look at Memorization in Deep Networks
We examine the role of memorization in deep learning, drawing connection...
research
∙
06/03/2016
Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations
We propose zoneout, a novel method for regularizing RNNs. At each timest...
research
∙
11/26/2015
Regularizing RNNs by Stabilizing Activations
We stabilize the activations of Recurrent Neural Networks (RNNs) by pena...
research
∙
02/13/2014