image-classification

分析了连续思维机（CTM）在ImageNet图像分类上的内部动力学，发现其思维轨迹是移动吸引不动点的曲线，注意力调度在图像间基本通用，表明CTM已找到一种处理所有ImageNet图像的统一策略。

该论文提出 SW-KAN，一种基于 Stieltjes-Wigert q-正交多项式的 Kolmogorov-Arnold 网络，通过指数-tanh 域映射和 O(N) 三递推计算，在图像分类与函数逼近任务中实现更优的精度-效率权衡。

PaLoRA introduces a paced low-rank adaptation method for LoRA-based continual learning, deriving an optimal adaptive pacing law that scales gradient step magnitude with the effective rank of accumulated knowledge. The method combines adaptive SVD truncation, null-space gradient projection, and rank-aware pacing, achieving consistent gains including 4% accuracy improvements on 50-task ImageNet-A and ImageNet-R benchmarks.

PSRben released official pretrained VisionHOPE-T/S/B visual backbone checkpoints on Hugging Face, covering ImageNet-1K classification, COCO detection/segmentation, and ADE20K semantic segmentation for the paper 'VisionHOPE: Visual Backbones as Self-Modifying Learning Systems'.

IntBMoE is a novel Mixture-of-Experts method that decouples participation, execution, and materialization costs through block-level conditioning and sparse execution, demonstrating improvements in image classification, language modeling, and recommendation systems, with real-world deployment in AMap's generative recommendation system.

An interactive game that challenges users to distinguish AI-generated images from real photographs within a 60-second time limit.

ALPINE introduces an ultra-lightweight spatial-relational architecture for few-shot image classification that achieves accuracy gains with fewer parameters, faster convergence, and better robustness compared to baselines like Prototypical Networks and MAML.

This paper proposes adaptive reciprocal knowledge distillation (AR-KD), a novel method that improves knowledge transfer from teacher to student models by simplifying the teacher's output distribution through relational alignment, achieving up to 7.13% accuracy gain on CIFAR-100 and ImageNet-1k datasets.

The paper introduces TAPe+MLv3, a compact computer vision system using structured representation for multi-task tasks, achieving competitive performance on benchmarks like COCO with fewer than 100,000 parameters.

The paper proposes CRAD, a class-wise reliability-aware distillation method for decentralized federated learning to handle heterogeneous architectures and non-IID data, achieving improved accuracy on image classification benchmarks.

This paper proposes FedCurv-DR, a lightweight federated continual learning method that reduces forgetting and balances performance, fairness, and energy efficiency for sustainable AI in cultural heritage applications.

The paper proposes Invertible Logits Transformation (InvLT), a post-hoc calibration method that applies a learnable scalar MLP element-wise to logits, preserving accuracy while correcting nonlinear miscalibration more flexibly than temperature scaling.

This paper presents the first systematic study of calibration under unseen subtype shift, showing that models become overconfident on novel subtypes within known coarse categories, and argues that subtype robustness should be evaluated with calibration metrics rather than accuracy alone.

This paper evaluates CNN pretraining strategies on the JONES-19 cultural design dataset, finding that learning from scratch with multi-crop sampling can match ImageNet pretraining for structured design data, suggesting smaller curated datasets may be more effective in specialized design domains.

SechKAN is a novel Kolmogorov-Arnold Network architecture that uses hyperbolic secant functions as basis functions, achieving competitive performance in function fitting, PDE problems, and image classification tasks while maintaining parameter efficiency comparable to MLPs.

This paper presents a novel parallel Quantum Convolutional Neural Network (QCNN) architecture that enables efficient classical simulation of large-scale QCNNs, demonstrating training on a 128-qubit model for MNIST binary classification.

Proposes a quantum-inspired hybrid classical-quantum framework for image classification using a mixture of experts, demonstrating improved performance and reduced failure rate on MNIST and Fashion-MNIST datasets.

This paper theoretically and empirically studies the relationship between counterfactual fairness (CF) and group fairness (GF) in image classification, introducing new CF evaluation datasets (CelebA-CF and LFW-CF). It finds that CF does not imply GF in images due to latent attributes correlated with sensitive attributes, and proposes Counterfactual Knowledge Distillation (CKD) to mitigate this.

This paper presents an empirical study comparing how different neural architectures (MLPs, CNNs, RNNs, pretrained transformers) degrade under temporal distribution shift across image and text domains, finding that models exploiting localized features degrade fastest while pretrained encoders drift more gradually.

This paper presents a scalable backpropagation-based algorithm for training deep convolutional networks to run on thermodynamic Ising hardware, achieving 94.9% on CIFAR-10 and 76.0% on CIFAR-100 while analyzing inference cost-accuracy tradeoffs.