Joar Max Viktor Skalse

Artificial Intelligence · 2019–2022

A researcher in the Science corpus: 3 papers; trusts 4 researchers and is trusted by 1 researcher here, as inferred from the citation graph.

Subfields

• Artificial Intelligence — 30 citations into it; #30,429 of 35,345 by trust here, #16,471 by citations
• Reinforcement Learning — 25 citations into it
• Alignment & RLHF — 12 citations into it

Whom Joar Max Viktor Skalse trusts overall

Their strongest outbound trust, strongest first, inferred from their citations.

1. Brian D. Ziebart
2. Andrew Y. Ng
3. Jan Leike
4. Tom Everitt
5. Nisan Stiennon
6. Richard S. Sutton
7. Dylan Hadfield-Menell
8. Geoffrey E. Hinton
9. David Krueger
10. Nick Bostrom

Who trusts Joar Max Viktor Skalse overall

1. Richard Ngo
2. Bram Wallace
3. Caglar Gulcehre
4. Stephen Casper

Papers from Joar Max Viktor Skalse's perspective

Ordered by how strongly the researchers Joar Max Viktor Skalse trusts endorse each paper. Same corpus, a different order for every researcher.

1. ImageNet classification with deep convolutional neural networks
2. Deep Residual Learning for Image Recognition (Computer Vision and Pattern Recognition)
3. Adam: A Method for Stochastic Optimization (International Conference on Learning Representations)
4. Attention is All you Need (Neural Information Processing Systems)
5. Gradient-based learning applied to document recognition
6. ImageNet: A large-scale hierarchical image database
7. Long Short-Term Memory
8. Sequence to Sequence Learning with Neural Networks (Neural Information Processing Systems)
9. Learning Multiple Layers of Features from Tiny Images
10. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift (International Conference on Machine Learning)

Their papers

• Risks from Learned Optimization in Advanced Machine Learning Systems (2019) — 297 citations
• Defining and Characterizing Reward Hacking (2022) — 169 citations
• Defining and Characterizing Reward Gaming (2022) — 379 citations