AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Non-Markovian Reward Modelling from Trajectory Labels via Interpretable Multiple Instance Learning

Joseph Early, Tom Bewley, Christine Evers, SArvapali Ramchurn

NeurIPS 2022 · Reinforcement learning › RL from feedback · approach: Theory of RLHF · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Replace Markovian Instance Space NN with Instance Space LSTMMIL reward model · Timerreturn MSE: 130.8 -> 7.313Improved
Replace Markovian Instance Space NN with CSC Instance Space LSTMMIL reward model · Movingreturn MSE: 22.24 -> 5.307Improved
Add concatenated skip connection to Instance Space LSTM (ablation)Instance Space LSTM · Chargerreturn MSE: 0.628 -> 0.125Improved
Add concatenated skip connection to Instance Space LSTM (ablation)Instance Space LSTM · Timerreward MSE: 0.263 -> 0.073Improved
Add concatenated skip connection to Instance Space LSTM (ablation)Instance Space LSTM · Lunar Landerreturn MSE: 1.223 -> 0.501Improved
Add concatenated skip connection to Instance Space LSTM (ablation)Instance Space LSTM · Keyreturn MSE: 0.488 -> 0.391Mixed
Use Embedding Space LSTM with post hoc reward computation instead of instance space rewardEmbedding Space LSTM · Timerreward MSE: 0.263 -> 101.8Hurt
Use Embedding Space LSTM instead of CSC Instance Space LSTMEmbedding Space LSTM · Lunar Landerreward MSE: 7.365 -> 120.2Hurt
Use Embedding Space LSTM instead of CSC Instance Space LSTMEmbedding Space LSTM · Lunar Landerreturn MSE: 0.501 -> 0.416Improved
Use CSC Instance Space LSTM reward model for RL agent training instead of oracle reward and hidden statesSAC / DQN agents · Timer and Lunar Landeroracle returnImproved
Train RL agents without hidden state informationRL agents with oracle reward · five RL tasksoracle returnHurt
Increase label swap noise from 0 to 0.5CSC Instance Space LSTM · five RL tasksreturn and reward predictionHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy