Brain Inspired Cortical Coding Method for Fast Clustering and Codebook Generation

Abstract

A major archetype of artificial intelligence is developing algorithms facilitating temporal efficiency and accuracy while boosting the generalization performance. Even with the latest developments in machine learning, a key limitation has been the inefficient feature extraction from the initial data, which is essential in performance optimization. Here, we introduce a feature extraction method inspired by energy–entropy relations of sensory cortical networks in the brain. Dubbed the brain-inspired cortex, the algorithm provides convergence to orthogonal features from streaming signals with superior computational efficiency while processing data in a compressed form. We demonstrate the performance of the new algorithm using artificially created complex data by comparing it with the commonly used traditional clustering algorithms, such as Birch, GMM, and K-means. While the data processing time is significantly reduced—seconds versus hours—encoding distortions remain essentially the same in the new algorithm, providing a basis for better generalization. Although we show herein the superior performance of the cortical coding model in clustering and vector quantization, it also provides potent implementation opportunities for machine learning fundamental components, such as reasoning, anomaly detection and classification in large scope applications, e.g., finance, cybersecurity, and healthcare.

High Performance Time Series Anomaly Detection Using Brain Inspired Cortical Coding Method

“…The cortical coding network is a self-organized, adaptive, hierarchical clustering based method. It has an n-array treelike structure and is a combination of transform coding and clustering [11]. The new method is inspired by the sensory cortical network found in biological brain, whose structuring is accomplished based on the rules of thermodynamics.…”

“…The spine nodes are the candidates that evolve to become cortex nodes, i.e., a well-trained spine node evolves into a cortex node. The detailed explanation of cortical coding network and its performance can be found in [11] and [37]. There are also other differences between cortex and spine nodes that define their structural characteristics and functions.…”

“…Although AI-based anomaly detection improvements partially compensate for this generalization drawback, their real-time applicability is limited due to the higher space and time complexity rates [9], [10]. Here we introduce a brain inspired cortical coding method that mimics the energy-entropy relationship of the neural network formation in the brain [11] and apply it to demonstrate time series anomaly detection in various data domains. The proposed cortical coding method is self organized, dynamic and adaptive.…”

“…The proposed cortical coding method is self organized, dynamic and adaptive. The key advantages of the proposed method are good generalization performance, low time complexity and the compliance with online learning [11].…”

“…Increasing the entropy at each level also increases the overall entropy at the leaves of the cortical coding tree. More detailed mathematical background of the entropy maximization in cortical coding network can be found in [11] and [37].…”