BioNIC: Biologically Inspired Neural Network for Image Classification Using Connectomics Principles

TL;DR

This work asks whether biologically constrained connectomics-inspired architectures can approach conventional neural networks on FER-2013 emotion recognition. BioNIC implements a four-layer cortical-column-inspired topology with inter- and intra-layer masks derived from MICrONs V1 data, integrated with Hebbian-like plasticity, Layer Normalization, data augmentation, synaptic noise, and graded inhibition. Ablation studies reveal that functional constraints (notably data augmentation and front-end convolutional processing) drive the majority of performance gains, while structural connectivity masks provide modest benefits, highlighting the value of combining biology-inspired learning rules with anatomical priors. The results demonstrate that connectome-constrained AI is computationally plausible, offering insights into generalization, robustness, and interpretability, and point to future work expanding to additional cortical areas and higher-order processing.

Abstract

We present BioNIC, a multi-layer feedforward neural network for emotion classification, inspired by detailed synaptic connectivity graphs from the MICrONs dataset. At a structural level, we incorporate architectural constraints derived from a single cortical column of the mouse Primary Visual Cortex(V1): connectivity imposed via adjacency masks, laminar organization, and graded inhibition representing inhibitory neurons. At the functional level, we implement biologically inspired learning: Hebbian synaptic plasticity with homeostatic regulation, Layer Normalization, data augmentation to model exposure to natural variability in sensory input, and synaptic noise to model neural stochasticity. We also include convolutional layers for spatial processing, mimicking retinotopic mapping. The model performance is evaluated on the Facial Emotion Recognition task FER-2013 and compared with a conventional baseline. Additionally, we investigate the impacts of each biological feature through a series of ablation experiments. While connectivity was limited to a single cortical column and biologically relevant connections, BioNIC achieved performance comparable to that of conventional models, with an accuracy of 59.77 $\pm$ 0.27% on FER-2013. Our findings demonstrate that integrating constraints derived from connectomics is a computationally plausible approach to developing biologically inspired artificial intelligence systems. This work also highlights the potential of new generation peta-scale connectomics data in advancing both neuroscience modeling and artificial intelligence.

Figures (17)

• Figure 1: Sample images from the FER-2013 dataset
• Figure 2: (Left) Nucleus detections highlighting one cortical column; (Right) Visualization of two neurons from different cortical layers.
• Figure 3: Model with biologically plausible inter-layer and intra-layer connections, graded inhibition, convolutional layers, and regularization
• Figure 4: Hypothesized layers in one cortical column
• Figure : Confusion Matrix