Aligning ML objectives with human values

Channel:
United States Simons Institute for the Theory of Computing
Subscribers:
68,700
Published on ● Video Link: https://www.youtube.com/watch?v=3fZvahTlPaQ


Duration: 47:00
750 views
11

Paul Christiano (Open AI)
https://simons.berkeley.edu/talks/tba-80
Emerging Challenges in Deep Learning



Other Videos By Simons Institute for the Theory of Computing


2019-08-07The Measure and Mismeasure of Fairness
2019-08-07Integrating Constraints into Deep Learning Architectures with Structured Layers
2019-08-07Towards Verified Deep Learning
2019-08-07Reinforcement Learning via an Optimization Lens
2019-08-06Reinforcement Learning in Recommender Systems: Some Challenges
2019-08-06Off-policy Policy Optimization
2019-08-06Optimality and Approximation with Policy Gradient Methods in Markov Decision Processes
2019-08-06On the Hardness of Reinforcement Learning With Value-function Approximation
2019-08-06Reinforcement Learning in Feature Space: Complexity and Regret
2019-08-05Flexible Neural Networks and the Frontiers of Meta-Learning
2019-08-05Aligning ML objectives with human values
2019-08-05Is Deeper Better Only When Shallow Is Good?
2019-08-05Efficient Deep Learning with Humans in the Loop
2019-08-05Machine Learning-based Design of Proteins and Small Molecules
2019-08-05Knowledge is embedded in language neural networks but can they reason?
2019-07-18Sample-complexity of Estimating Convolutional and Recurrent Neural Networks
2019-07-18Studying Generalization in Deep Learning via PAC-Bayes
2019-07-18Towards Understanding Transfer Learning with Applications to Medical Imaging
2019-07-18Kernel and Deep Regimes in Overparameterized Learning
2019-07-18Meta-learning of Optimizers and Update Rules
2019-07-18Robust Deep Learning Under Distribution Shift


Tags:
Emerging Challenges in Deep Learning
Simons Institute
Theory of Computing
Theory of Computation
Theoretical Computer Science
Computer Science
UC Berkeley
Paul Christiano