Research map / Probabilistic and scientific modeling / Variational and generative sampling
Competing Mutual Information Constraints with Stochastic Competition-Based Activations for Learning Diversified Representations
Konstantinos P. Panousis, Anastasios Antoniadis, Sotirios Chatzis
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 ReLU activations with stochastic LWTA-2 and use IBP utility inference | ICP · CIFAR-10, VGG-16 | error (%): 6.10 -> 6.01 | Improved |
| replace ReLU activations with stochastic LWTA-2 and use IBP utility inference | ICP · CIFAR-10, GoogLeNet | error (%): 4.26 -> 4.31 | Hurt |
| replace ReLU activations with stochastic LWTA-2 and use IBP utility inference | ICP · CIFAR-10, ResNet20 | error (%): 6.01 -> 5.94 | Improved |
| replace ReLU activations with stochastic LWTA-2 and use IBP utility inference | ICP · CIFAR-10, DenseNet40 | error (%): 4.99 -> 4.78 | Improved |
| replace ReLU activations with stochastic LWTA-2 and use IBP utility inference | ICP · CIFAR-100, VGG-16 | error (%): 24.54 -> 24.35 | Improved |
| replace ReLU activations with stochastic LWTA-2 and use IBP utility inference | ICP · CIFAR-100, GoogLeNet | error (%): 18.55 -> 19.00 | Hurt |
| replace ReLU activations with stochastic LWTA-2 and use IBP utility inference | ICP · CIFAR-100, ResNet20 | error (%): 28.13 -> 28.02 | Improved |
| replace ReLU activations with stochastic LWTA-2 and use IBP utility inference | ICP · CIFAR-100, DenseNet40 | error (%): 24.52 -> 24.44 | Improved |
| add stochastic rather than deterministic maximum-activation competition (ablation) | ICP LWTA-2 · CIFAR-10, VGG-16 | error (%): 6.34 -> 6.23 | Improved |
| add stochastic rather than deterministic maximum-activation competition (ablation) | ICP IBP & LWTA-2 · CIFAR-10, VGG-16 | error (%): 6.27 -> 6.01 | Improved |
| add IBP component-utility inference (ablation) | ICP LWTA-2 · CIFAR-10, VGG-16 | error (%): 6.23 -> 6.01 | Improved |
| add stochastic rather than deterministic maximum-activation competition (ablation) | ICP LWTA-4 · CIFAR-10, VGG-16 | error (%): 7.01 -> 6.85 | Improved |
Closest papers by meaning
- Convolutional Channel-Wise Competitive Learning for the Forward-Forward Algorithm (AAAI 2024 · similarity 0.85)
- When Representations Align: Universality in Representation Learning Dynamics (ICML 2024 · similarity 0.84)
- Global Convergence and Rich Feature Learning in $L$-Layer Infinite-Width Neural Networks under $\mu$ Parametrization (ICML 2025 · similarity 0.84)
- From Flat to Hierarchical: Extracting Sparse Representations with Matching Pursuit (NeurIPS 2025 · similarity 0.83)
- Bridging Critical Gaps in Convergent Learning: How Representational Alignment Evolves Across Layers, Training, and Distribution Shifts (NeurIPS 2025 · similarity 0.83)
- Why DiT Models Underperform as Representation Learners without Long Skip Connections (NeurIPS 2026 · similarity 0.83)
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.
