cifar-10
A widely used dataset in machine learning that consists of 60,000 32x32 color images in 10 different classes, with 6,000 images per class. It serves as a standard benchmark for evaluating image classification algorithms.
- Adaptive Discretization for Consistency Models
- Adaptive Stochastic Coefficients for Accelerating Diffusion Sampling
- Adversary Aware Optimization for Robust Defense
- Beyond Masked and Unmasked: Discrete Diffusion Models via Partial Masking
- Beyond Random: Automatic Inner-loop Optimization in Dataset Distillation
- Efficient Multi-bit Quantization Network Training via Weight Bias Correction and Bit-wise Coreset Sampling
- Efficient and Generalizable Mixed-Precision Quantization via Topological Entropy
- Energy Matching: Unifying Flow Matching and Energy-Based Models for Generative Modeling
- FedFree: Breaking Knowledge-sharing Barriers through Layer-wise Alignment in Heterogeneous Federated Learning
- Knowledge Distillation Detection for Open-weights Models
- Learning to Integrate Diffusion ODEs by Averaging the Derivatives
- Mind the Gap: Removing the Discretization Gap in Differentiable Logic Gate Networks
- ModHiFi: Identifying High Fidelity predictive components for Model Modification
- TransferBench: Benchmarking Ensemble-based Black-box Transfer Attacks
- Variational Supervised Contrastive Learning
- Vulnerable Data-Aware Adversarial Training
- When Does Closeness in Distribution Imply Representational Similarity? An Identifiability Perspective