Brain Cognition-Inspired Dual-Pathway CNN Architecture for Image Classification

• Published in: IEEE transaction on neural networks and learning systems 2024-07, Vol.35 (7), p.9900-9914
• Main Authors: Dong, Songlin, Gong, Yihong, Shi, Jingang, Shang, Miao, Tao, Xiaoyu, Wei, Xing, Hong, Xiaopeng, Zhou, Tiangang
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Brain architecture; Cognition; Cognition inspired; cognition-inspired network (CogNet); convolutional neural network (CNN); Convolutional neural networks; dual pathway; Feature extraction; Image classification; Information processing; Information processing (biology); Modulation; Neural networks; Neuroimaging; semantic confusion (SC); Semantics; State-of-the-art reviews; texture bias (TB); transformer; Transformers; Visual system; Visualization
• ISSN: 2162-237X; 2162-2388; 2162-2388
• DOI: 10.1109/TNNLS.2023.3237962

Inspired by the global-local information processing mechanism in the human visual system, we propose a novel convolutional neural network (CNN) architecture named cognition-inspired network (CogNet) that consists of a global pathway, a local pathway, and a top-down modulator. We first use a common CNN block to form the local pathway that aims to extract fine local features of the input image. Then, we use a transformer encoder to form the global pathway to capture global structural and contextual information among local parts in the input image. Finally, we construct the learnable top-down modulator where fine local features of the local pathway are modulated by global representations of the global pathway. For ease of use, we encapsulate the dual-pathway computation and modulation process into a building block, called the global-local block (GL block), and a CogNet of any depth can be constructed by stacking a necessary number of GL blocks one after another. Extensive experimental evaluations have revealed that the proposed CogNets have achieved the state-of-the-art performance accuracies on all the six benchmark datasets and are very effective for overcoming the "texture bias" and the "semantic confusion" problems faced by many CNN models.