Research map / Adversarial robustness and security / Attacks on text and graphs
Improving Neural Language Modeling via Adversarial Training
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 adversarial softmax training to AWD-LSTM + MoS + Partial Shuffled | AWD-LSTM + MoS + Partial Shuffled · PTB test | perplexity: 47.49 -> 46.01 | Helped |
| Add adversarial softmax training to AWD-LSTM | AWD-LSTM · WT2 test | perplexity: 44.30 -> 40.71 | Helped |
| Add adversarial training to 4 layer QRNN | 4 layer QRNN · WT103 test | perplexity: 33.0 -> 31.6 | Helped |
| Add adversarial training to Transformer Base | Transformer Base · WMT2014 En→De test | BLEU: 27.30 -> 28.43 | Helped |
| Add adversarial training to Transformer Small | Transformer Small · IWSLT2014 De→En test | BLEU: 32.47 -> 33.61 | Helped |
| Use adversarial perturbation magnitude α in [0.001,0.01] rather than zero (ablation) | PTB language model · PTB | perplexity | Helped |
| Increase perturbation magnitude to α=0.05 (ablation) | PTB language model · PTB | perplexity | Hurt |
| Back-propagate through the ε||h|| norm term (ablation) | Adversarial MLE · language modeling | training error | Hurt |
Closest papers by meaning
- Adversarial Training for Improving Model Robustness? Look at Both Prediction and Interpretation (AAAI 2022 · similarity 0.89)
- Toward Adversarial Training on Contextualized Language Representation (ICLR 2023 · similarity 0.89)
- Enhancing NLU in Large Language Models Using Adversarial Noisy Instruction Tuning (AAAI 2025 · similarity 0.88)
- Efficient Adversarial Training in LLMs with Continuous Attacks (NeurIPS 2024 · similarity 0.86)
- Confidence Elicitation: A New Attack Vector for Large Language Models (ICLR 2025 · similarity 0.86)
- Benchmarking Optimizers for Large Language Model Pretraining (NeurIPS 2026 · 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.