Research map / Label-efficient and robust learning / Knowledge distillation
MTSS: Learn from Multiple Domain Teachers and Become a Multi-Domain Dialogue Expert
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 |
|---|---|---|---|
| Train HRED with multiple domain-specific teachers and both text- and policy-level guidance | HRED student · MultiWOZ multi-domain | Inform: 70.7 -> 77.5 | Helped |
| Train HRED with multiple domain-specific teachers and both text- and policy-level guidance | HRED student · MultiWOZ multi-domain | Success: 60.9 -> 64.9 | Helped |
| Train HRED with multiple domain-specific teachers and both text- and policy-level guidance | HRED student · MultiWOZ multi-domain | BLEU: 17.5 -> 18.7 | Helped |
| Use MTSS instead of HRED + TRADE | HRED-based dialogue system · MultiWOZ Restaurant | Success: 74.4 -> 81.2 | Helped |
| Use MTSS instead of HRED + TRADE | HRED-based dialogue system · MultiWOZ Hotel | Success: 50.5 -> 81.5 | Mixed |
| Use MTSS instead of HDSA + TRADE | dialogue system · MultiWOZ multi-domain | BLEU: 20.1 -> 18.7 | Hurt |
| Use MTSS instead of HDSA + TRADE | dialogue system · MultiWOZ multi-domain | Inform: 76.4 -> 77.5 | Hurt |
| Add both text-level and policy-level distillation (ablation) | HRED student · MultiWOZ multi-domain | Inform: 70.7 -> 77.5 | Helped |
| Add text-level distillation alone (α1=0.005) (ablation) | HRED student · MultiWOZ multi-domain | Success: 60.9 -> 63.4 | Helped |
| Add policy-level distillation alone (α2=0.005) (ablation) | HRED student · MultiWOZ multi-domain | Inform: 70.7 -> 74.6 | Helped |
| Use both guidance losses rather than text-level distillation alone (ablation) | HRED student · MultiWOZ multi-domain | Inform: 72.2 -> 77.5 | Helped |
| Use both guidance losses rather than policy-level distillation alone (ablation) | HRED student · MultiWOZ multi-domain | Success: 63.2 -> 64.9 | Helped |
Closest papers by meaning
- A Student-Teacher Architecture for Dialog Domain Adaptation Under the Meta-Learning Setting (AAAI 2021 · similarity 0.88)
- Multi-View Feature Representation for Dialogue Generation with Bidirectional Distillation (AAAI 2021 · similarity 0.87)
- Learning from Easy to Complex: Adaptive Multi-Curricula Learning for Neural Dialogue Generation (AAAI 2020 · similarity 0.86)
- Improving Open-Domain Dialogue Response Generation with Multi-Source Multilingual Commonsense Knowledge (AAAI 2024 · similarity 0.85)
- StateTree: Enhancing Long-term Dialogue Reasoning via Reinforcement Learning (NeurIPS 2026 · similarity 0.85)
- ChatterBox: Multimodal Referring and Grounding with Chain-of-Questions (AAAI 2025 · similarity 0.84)
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.