V-EfficientNets: Vector-Valued Efficiently Scaled Convolutional Neural Network Models

2025-05-08

AI Summary

Existing convolutional neural networks lack native support for multidimensional vector-valued data—such as channel-correlated features in medical images—limiting geometric structure preservation and channel-wise relational modeling. Method: We propose V-EfficientNet, the first scalable convolutional network explicitly designed for vector-valued inputs. It introduces a vector convolution operator preserving geometric structure, integrates vector neural network principles into compound scaling for the first time, and incorporates channel-coupled normalization and a lightweight vector attention module to explicitly model intrinsic geometric relationships among channels. Results: On the ALL-IDB2 acute lymphoblastic leukemia classification task, V-EfficientNet achieves a mean accuracy of 99.46%, with significantly fewer parameters than EfficientNet-B0–B7. It demonstrates simultaneous gains in both accuracy and efficiency, establishing a novel paradigm for vector-valued medical image analysis.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Matrix & Tensor Methods
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

EfficientNet models are convolutional neural networks optimized for parameter allocation by jointly balancing network width, depth, and resolution. Renowned for their exceptional accuracy, these models have become a standard for image classification tasks across diverse computer vision benchmarks. While traditional neural networks learn correlations between feature channels during training, vector-valued neural networks inherently treat multidimensional data as coherent entities, taking for granted the inter-channel relationships. This paper introduces vector-valued EfficientNets (V-EfficientNets), a novel extension of EfficientNet designed to process arbitrary vector-valued data. The proposed models are evaluated on a medical image classification task, achieving an average accuracy of 99.46% on the ALL-IDB2 dataset for detecting acute lymphoblastic leukemia. V-EfficientNets demonstrate remarkable efficiency, significantly reducing parameters while outperforming state-of-the-art models, including the original EfficientNet. The source code is available at https://github.com/mevalle/v-nets.

Problem

• Extends EfficientNet to handle vector-valued data efficiently
• Improves accuracy in medical image classification tasks
• Reduces parameters while outperforming state-of-the-art models

Innovation

• Vector-valued EfficientNet extension for multidimensional data
• Joint optimization of width, depth, and resolution
• Reduced parameters with higher accuracy performance