Machine learning approaches are gaining popularity in the medical field for diagnostics, predictive analytics and general research. With data often being unlabeled or sparse to collect, there is a need for unsupervised learning networks in the medical field. Self-Organizing Feature Maps (SOFM) are a common application of unsupervised networks and allow for the use of unlabeled data in their training. We applied chest x-ray images of COVID-19 patients to an SOFM network and found a distinct classification between sick and healthy patients with an average euclidean distance of 1.1 between 1st and 2nd winning neurons in our testing set. We were also able to show which features of the input space had the highest weight on the classification, to study saliency of features on this unsupervised network. This work shows that unsupervised learning is able to extract features of medical data, specifically chest x-rays of COVID-19 patients, while also accurately classifying the image. This SOFM network can be found at https://github.com/king2b3/SOFM.

Many currently available hardware implementations of the unsupervised self-organizing feature map (SOFM) algorithm utilize CMOS-only circuits that often compromise key behaviors of the SOFM algorithm due to complexity. We propose a neuromorphic architecture harnessing the unique properties of FeFETs and gated-RRAM for in-memory computing to implement the SOFM algorithm. The FeFET-based synapse, organized in a novel circuit, is able to compute the input-weight Euclidean error in memory via the saturation drain current. The selfdecaying states of the gated-RRAM allow for a self-decaying neighborhood and learning rate implementation to allow for convergence and lifelong learning. This novel architecture is able to successfully cluster benchmarks (RGB colors and MNIST handwritten digits) and real-life datasets such as COVID-19 patient chest x-rays completely unsupervised. The architecture also demonstrates a significant amount of robustness to device variability and damaged neurons. Additionally, the proposed architecture is completely parallelized and provides a power efficient platform for implementing the SOFM algorithm.

No abstract

No abstract

No abstract