Research map / Large language models / Retrieval and knowledge
Dialog-based Language Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| add reward-based imitation + forward prediction jointly (RBI+FP) vs single strategies | MemN2N · bAbI tasks 1-10 | number of completed tasks (>=95%): 10 vs 9 (RBI) and 5 (FP) at pi_acc=0.5 | Helped |
| use forward prediction instead of imitation when teacher gives feedback (Tasks 3-5) | MemN2N · MovieQA, pi_acc=0.5 | test accuracy (%): 60 (FP) vs 48 (imitation) on Task 3 | Helped |
| forward prediction without any reward signal | MemN2N · bAbI Task 7 (No Feedback), pi_acc=0.5 | test accuracy (%): 100 (FP) vs 20 (RBI) | Helped |
| reward-based imitation vs imitation learning on bAbI expert-student task (Task 1) | MemN2N · bAbI Task 1, pi_acc=0.5 | test accuracy (%): 100 (RBI) vs 100 (imitation) — no difference at pi_acc=0.5; FP only 23 | Helped |
| forward prediction on bAbI Task 2 (positive/negative feedback only, no rewards) vs reward-based imitation | MemN2N · bAbI Task 2, pi_acc=0.5 | test accuracy (%): 93 (FP) vs 99 (RBI) | Hurt |
| biased policy in Task 2 (fixed 'bathroom' answer half the time when wrong) (ablation) | MemN2N forward prediction · bAbI Task 2 | test accuracy (%): 68% vs 93% under the standard policy at pi_acc=0.5 | Hurt |
| reward-based imitation with sparse rewards (pi_acc from 0.5 to 0.01) (ablation) | MemN2N RBI · bAbI Task 2 | test accuracy (%): 99 at pi_acc=0.5 vs 91 at pi_acc=0.01 | Hurt |
| forward prediction vs imitation when policy is inaccurate (MovieQA) | MemN2N · MovieQA Task 3, pi_acc=0.01 | test accuracy (%): 58 (FP) vs 26 (imitation) | Helped |
Closest papers by meaning
- Reasoning in Dialog: Improving Response Generation by Context Reading Comprehension (AAAI 2021 · similarity 0.89)
- Exploring Auxiliary Reasoning Tasks for Task-oriented Dialog Systems with Meta Cooperative Learning (AAAI 2021 · similarity 0.87)
- Two Can Play This Game: Visual Dialog With Discriminative Question Generation and Answering (CVPR 2018 · similarity 0.87)
- An Efficient Self-Learning Framework For Interactive Spoken Dialog Systems (ICML 2024 · similarity 0.87)
- Learning to Clarify: Multi-turn Conversations with Action-Based Contrastive Self-Training (ICLR 2025 · similarity 0.86)
- TA&AT: Enhancing Task-Oriented Dialog with Turn-Level Auxiliary Tasks and Action-Tree Based Scheduled Sampling (AAAI 2024 · similarity 0.86)
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.