Research map / Architectures and efficiency / Attention and transformers
Self-attention with Functional Time Representation Learning
Da Xu, Chuanwei Ruan, Evren Korpeoglu, Sushant Kumar, Kannan Achan
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace positional encoding with Mercer time embedding (ablation) | self-attention · Stack Overflow | accuracy: 44.03 -> 46.83 | Improved |
| replace positional encoding with Mercer time embedding (ablation) | self-attention · MovieLens-1m | NDCG@10: 59.05 -> 61.67 | Improved |
| replace positional encoding with Mercer time embedding (ablation) | self-attention · Walmart.com | Hit@10: 12.49 -> 14.94 | Improved |
| replace positional encoding with Bochner Non-para time embedding (ablation) | self-attention · Walmart.com | Hit@10: 12.49 -> 13.16 | Improved |
| replace positional encoding with Bochner Normal time embedding (ablation) | self-attention · Walmart.com | Hit@10: 12.49 -> 7.66 | Hurt |
| replace positional encoding with Bochner Normal time embedding (ablation) | self-attention · Stack Overflow | accuracy: 44.03 -> 44.89 | No clear effect |
| replace positional encoding with Bochner Inv CDF time embedding (ablation) | self-attention · Stack Overflow | accuracy: 44.03 -> 44.67 | No clear effect |
| replace positional encoding with Bochner Inv CDF time embedding (ablation) | self-attention · Walmart.com | Hit@10: 12.49 -> 12.77 | No clear effect |
| replace positional encoding with Mercer time embedding (ablation) | self-attention · MovieLens-1m | Hit@10: 82.45 -> 82.92 | No clear effect |
| use larger Fourier degree k for Mercer embedding at fixed d (ablation) | Mercer time embedding · MovieLens-1m | Hit@10 | Hurt |
Closest papers by meaning
- Self-Attentive Hawkes Process (ICML 2020 · similarity 0.87)
- Learning Sequence Representations by Non-local Recurrent Neural Memory (IJCV 2022 · similarity 0.87)
- Context-Aware Self-Attention Networks (AAAI 2019 · similarity 0.86)
- Staircase Attention for Recurrent Processing of Sequences (NeurIPS 2022 · similarity 0.86)
- Kernel Representation Learning for Time Sequence: Algorithm, Theory, and Applications (AAAI 2025 · similarity 0.84)
- From Self-Attention to Markov Models: Unveiling the Dynamics of Generative Transformers (ICML 2024 · 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.
