Abstract:Deep neural networks (DNNs) and, in particular, convolutional neural networks (CNNs) have brought significant advances in a wide range of modern computer application problems. However, the increasing availability of large amounts of datasets as well as the increasing available computational power of modern computers lead to a steady growth in the complexity and size of DNN and CNN models, respectively, and thus, to longer training times. Hence, various methods and attempts have been developed to accelerate and parallelize the training of complex network architectures. In this work, a novel CNN-DNN architecture is proposed that naturally supports a model parallel training strategy and that is loosely inspired by two-level domain decomposition methods (DDM). First, local CNN models, that is, subnetworks, are defined that operate on overlapping or nonoverlapping parts of the input data, for example, sub-images. The subnetworks can be trained completely in parallel and independently of each other. Each subnetwork then outputs a local decision for the given machine learning problem which is exclusively based on the respective local input data. Subsequently, in a second step, an additional DNN model is trained which evaluates the local decisions of the local subnetworks and generates a final, global decision. In this paper, we apply the proposed architecture to image classification problems using CNNs. Experimental results for different 2D image classification problems are provided as well as a face recognition problem, and a classification problem for 3D computer tomography (CT) scans. Therefore, classical ResNet and VGG architectures are considered. The results show that the proposed approach can significantly accelerate the required training time compared to the global model and, additionally, can also help to improve the accuracy of the underlying classification problem.

Explaining NonLinear Classification Decisions with Deep Taylor Decomposition

Decompose Learning: Combine Feature Extraction and Classification

Local Interpretation of Transformer Based on Linear Decomposition

DXAI: Explaining Classification by Image Decomposition

Towards Explaining Anomalies: A Deep Taylor Decomposition of One-Class Models

Deeper Interpretability of Deep Networks

DecompX: Explaining Transformers Decisions by Propagating Token Decomposition

How to Explain Neural Networks: A perspective of data space division

Taylor Convolutional Networks for Image Classification

Domain-decomposed image classification algorithms using linear discriminant analysis and convolutional neural networks

Take 5: Interpretable Image Classification with a Handful of Features

M-Rule: an Enhanced Deep Taylor Decomposition for Multi-model Interpretability

NeuroView: Explainable Deep Network Decision Making

A Domain Decomposition-Based CNN-DNN Architecture for Model Parallel Training Applied to Image Recognition Problems

Knowledge Discovery from Layered Neural Networks based on Non-negative Task Decomposition

On the Interpretation of Convolutional Neural Networks for Text Classification

Interpreting Deep Learning Model Using Rule-based Method

Deeply Explain CNN Via Hierarchical Decomposition

Interpretable Neural Network Decoupling.

Decom--CAM: Tell Me What You See, In Details! Feature-Level Interpretation via Decomposition Class Activation Map

Interpreting the decisions of CNNs via influence functions