Diffusion-Based Feature Denoising with NNMF for Robust handwritten digit multi-class classification

arXiv cs.CV / 4/1/2026

💬 OpinionSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper proposes a robust handwritten digit multi-class classification framework that performs diffusion-based feature denoising in feature space to resist noise and adversarial attacks.
It first transforms input images into interpretable, nonnegative parts using Nonnegative Matrix Factorization (NNMF), while also extracting deep features with a CNN.
The method combines NNMF-derived features and CNN features into a hybrid representation, then applies a step-wise diffusion process by adding Gaussian noise to the feature representation.
A dedicated feature denoiser network is trained to reverse the diffusion/noising process and reconstruct cleaner feature representations before classification.
Experiments, including adversarial evaluation with AutoAttack, show the diffusion-based hybrid model maintains strong performance and demonstrates improved robustness compared with CNN baselines.

Abstract

This work presents a robust multi-class classification framework for handwritten digits that combines diffusion-driven feature denoising with a hybrid feature representation. Inspired by our previous work on brain tumor classification, the proposed approach operates in a feature space to improve the robustness to noise and adversarial attacks. First, the input images are converted into tight, interpretable exemplification using Nonnegative Matrix Factorization (NNMF). In parallel, special deep features are extracted using a computational neural network (CNN). These integral features are combined into a united hybrid representation. To improve robustness, a step diffusion operation is used in the feature space by gradually adding Gaussian noise. A feature denoiser network is trained to reverse this operation and rebuild clean representations from tilted inputs. The courteous features are then applied for multi-class classification. The suggested method is evaluated in both baseline and adversarial settings using AutoAttack. The experimental outcome present that the diffusion-based hybrid model is both effective and robust, the CNN baseline models outperforming while maintain powerful classification performance. These results explain the activity of feature-level diffusion defense for reliable multi-class handwritten digit classification.