Neuromorphic Computing for Content-based Image Retrieval

TL;DR

This paper demonstrates that neuromorphic computing using Intel's Loihi chip can perform content-based image retrieval with significantly lower energy consumption while maintaining accuracy, highlighting its potential for low-power visual search applications.

Contribution

The study evaluates Loihi's application to image retrieval, showing it achieves high energy efficiency compared to traditional CPUs and GPUs without sacrificing accuracy.

Findings

Neuromorphic solution is 2.5x more energy-efficient than ARM Cortex-A72.

Neuromorphic solution is 12.5x more energy-efficient than NVIDIA T4 GPU.

Maintains the same matching accuracy as traditional methods.

Abstract

Neuromorphic computing mimics the neural activity of the brain through emulating spiking neural networks. In numerous machine learning tasks, neuromorphic chips are expected to provide superior solutions in terms of cost and power efficiency. Here, we explore the application of Loihi, a neuromorphic computing chip developed by Intel, for the computer vision task of image retrieval. We evaluated the functionalities and the performance metrics that are critical in content-based visual search and recommender systems using deep-learning embeddings. Our results show that the neuromorphic solution is about 2.5 times more energy-efficient compared with an ARM Cortex-A72 CPU and 12.5 times more energy-efficient compared with NVIDIA T4 GPU for inference by a lightweight convolutional neural network without batching while maintaining the same level of matching accuracy. The study validates the potential of neuromorphic computing in low-power image retrieval, as a complementary paradigm to the existing von Neumann architectures.