Research map / Architectures and efficiency / RNNs and normalization
LassoFlexNet: a Flexible Neural Architecture for Tabular Data
Kry Yik-Chau Lui, Cheng Chi, Kishore Basu, Yanshuai Cao
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 |
|---|---|---|---|
| add EMA to Seq-Hier-Prox (ablation) | LassoFlexNet · California Housing | test RMSE: 0.4195 -> 0.4047 | Improved |
| replace LassoNet Hier-Prox with sequential Seq-Hier-Prox (ablation) | LassoFlexNet · California Housing | test RMSE: 0.4320 -> 0.4047 | Improved |
| add proximal gradient (vs Adam without proximal) (ablation) | LassoFlexNet · California Housing | test RMSE: 0.4272 -> 0.4047 | Improved |
| add piecewise linear encoding (PLE) (ablation) | LassoFlexNet · California Housing | test RMSE: 0.4387 -> 0.4047 | Improved |
| add per-feature embeddings (PFE) (ablation) | LassoFlexNet · California Housing | test RMSE: 0.5371 -> 0.4047 | Improved |
| use nonlinear instead of linear PFE layers (ablation) | LassoFlexNet · California Housing | test RMSE: 0.4145 -> 0.4047 | Improved |
| use parameter-free BatchNorm instead of learnable BatchNorm (ablation) | LassoFlexNet · California Housing | test RMSE: 0.4191 -> 0.4047 | Improved |
| use MLP-Mixer instead of a plain MLP backbone (ablation) | LassoFlexNet · California Housing | test RMSE: 0.8453 -> 0.4047 (MLP 1); 0.8406 -> 0.4047 (MLP 2) | Improved |
| add linear skip connection to the Mixer (ablation) | LassoFlexNet · California Housing | test RMSE: 0.4221 -> 0.4047 | Improved |
| use tau < 1 soft curriculum (vs tau = 1) (ablation) | LassoFlexNet · California Housing | test RMSE: 0.4047 -> 0.4004 (tau = 1) | No clear effect |
| add PLE to linear LassoNet (ablation) | Linear LassoNet · California Housing | test RMSE: 0.5263 -> 0.5226 | No clear effect |
| switch LassoNet optimizer to Adam or Adam with HGD (ablation) | Linear LassoNet · California Housing | test RMSE: 0.5263 -> 0.5291 (Adam), 0.5294 (AdamHGD) | No clear effect |
Closest papers by meaning
- TabArena: A Living Benchmark for Machine Learning on Tabular Data (NeurIPS 2025 · similarity 0.88)
- TabM: Advancing tabular deep learning with parameter-efficient ensembling (ICLR 2025 · similarity 0.88)
- Hybrid Autoencoders for Tabular Data: Leveraging Model-Based Augmentation in Low-Label Settings (NeurIPS 2025 · similarity 0.88)
- TabFlex: Scaling Tabular Learning to Millions with Linear Attention (ICML 2025 · similarity 0.87)
- BiSHop: Bi-Directional Cellular Learning for Tabular Data with Generalized Sparse Modern Hopfield Model (ICML 2024 · similarity 0.87)
- Retrieval & Fine-Tuning for In-Context Tabular Models (NeurIPS 2024 · similarity 0.87)
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.
