Research map / Learning theory and algorithms / Prediction and decision losses
Learning Energy Networks with Generalized Fenchel-Young Losses
Mathieu Blondel, Felipe Llinares-López, Robert Dadashi, Leonard Hussenot, Matthieu Geist
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 pairwise label interactions instead of using the unary linear model | Generalized Fenchel-Young energy networks · yeast | test accuracy (%): 79.76 -> 80.19 | Improved |
| Add pairwise label interactions instead of using the unary linear model | Generalized Fenchel-Young energy networks · scene | test accuracy (%): 89.14 -> 91.58 | Improved |
| Add pairwise label interactions instead of using the unary linear model | Generalized Fenchel-Young energy networks · mediamill | test accuracy (%): 96.84 -> 96.95 | Improved |
| Add pairwise label interactions instead of using the unary linear model | Generalized Fenchel-Young energy networks · birds | test accuracy (%): 86.47 -> 91.55 | Improved |
| Add pairwise label interactions instead of using the unary linear model | Generalized Fenchel-Young energy networks · emotions | test accuracy (%): 78.22 -> 80.56 | Improved |
| Add pairwise label interactions instead of using the unary linear model | Generalized Fenchel-Young energy networks · cal500 | test accuracy (%): 85.67 -> 85.73 | Improved |
| Make the SPEN prior energy concave in the output | SPEN · yeast | test accuracy (%): 79.99 -> 80.00 | No clear effect |
| Make the SPEN prior energy concave in the output | SPEN · scene | test accuracy (%): 91.24 -> 90.64 | Hurt |
| Make the SPEN prior energy concave in the output | SPEN · mediamill | test accuracy (%): 96.68 -> 96.95 | Improved |
| Make the SPEN prior energy concave in the output | SPEN · birds | test accuracy (%): 91.41 -> 91.77 | Improved |
| Make the SPEN prior energy concave in the output | SPEN · emotions | test accuracy (%): 79.35 -> 79.73 | Improved |
| Make the SPEN prior energy concave in the output | SPEN · cal500 | test accuracy (%): 86.25 -> 86.35 | Improved |
Closest papers by meaning
- Joint Learning of Energy-based Models and their Partition Function (ICML 2025 · similarity 0.87)
- Learning with Fitzpatrick Losses (NeurIPS 2024 · similarity 0.85)
- Learning Energy Decompositions for Partial Inference in GFlowNets (ICLR 2024 · similarity 0.85)
- Energy Loss Functions for Physical Systems (NeurIPS 2025 · similarity 0.84)
- Only Strict Saddles in the Energy Landscape of Predictive Coding Networks? (NeurIPS 2024 · similarity 0.84)
- A Theoretical Framework For Overfitting In Energy-based Modeling (ICML 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.
