Learning Spiking Neural Network Models of Drosophila Olfaction

Published in International Conference on Neuromorphic Systems, ICONS, 2020

We present research in the modeling of neurons within Drosophila (fruit fly) olfaction. We describe the process from data collection, to model creation, and spike generation. Our approach utilizes com- putational elements such as spiking neural networks that employ leaky integrate-and-fire neurons with adaptive firing behavior that more closely mimick biological neurons. We describe the methods of several learning implementations in both software and hard- ware. Finally, we present both quantitative and qualitative results on learning spiking neural network models.

Download here

Modeling Biological Immunity to Adversarial Examples

Published in IEEE Computer Vision and Pattern Recognition, CVPR, 2020

In a general sense, adversarial attack through perturbations is not a machine learning vulnerability. Human and biological vision can also be fooled by various methods, i.e. mixing high and low frequency images together, by altering semantically related signals, or by sufficiently distorting the input signal. However, the amount and magnitude of such a distortion required to alter biological perception is at a much larger scale. In this work, we explored this gap through the lens of biology and neuroscience in order to understand the robustness exhibited in human perception. Our experiments show that by leveraging sparsity and modeling the biological mechanisms at a cellular level, we are able to mitigate the effect of adversarial alterations to the signal that have no perceptible meaning. Furthermore, we present and illustrate the effects of top-down functional processes that contribute to the inherent immunity in human perception in the context of exploiting these properties to make a more robust machine vision system.

Download here

Using Sinusoidally-Modulated Noise as a Surrogate for Slow-Wave Sleep to Accomplish Stable Unsupervised Dictionary Learning in a Spike-Based Sparse Coding Models

Published in IEEE Computer Vision and Pattern Recognition Workshops, CVPRW, 2020

Sparse coding algorithms have been used to model the acquisition of V1 simple cell receptive fields as well as to accomplish the unsupervised acquisition of features for a variety of machine learning applications. The Locally Com- petitive Algorithm (LCA) provides a biologically plausible implementation of sparse coding based on lateral inhibi- tion. LCA can be reformulated to support dictionary learn- ing via an online local Hebbian rule that reduces predictive coding error. Although originally formulated in terms of leaky integrator rate-coded neurons, LCA based on lateral inhibition between leaky integrate-and-fire (LIF) neurons has been implemented on spiking neuromorphic processors but such implementations preclude local online learning. We previously reported that spiking LCA can be expressed in terms of predictive coding error in a manner that allows for unsupervised dictionary learning via a local Hebbian rule but the issue of stability has not previously been ad- dressed.

Download here

Guided Ultrasound Imaging using a Deep Regression Network

Published in Ultrasonic Imaging and Tomography, SPIE Medical Imaging, 2020

In this work, we present a machine learning method to guide an ultrasound operator towards a selected area of interest. Unlike other automatic medical imaging methods, ultrasound imaging is one of the few imaging modalities where the operator’s skill and training are critical in obtaining high quality images. Additionally, due to recent advances in affordability and portability of ultrasound technology, its utilization by non-experts has increased. Thus, there is a growing need for intelligent systems that have the ability to assist ultrasound operators in both clinical and non-clinical scenarios. We propose a system that leverages machine learning to map real time ultrasound scans to transformation vectors that can guide a user to a target organ or anatomical structure. We present a unique training system that passively collects supervised training data from an expert sonographer and uses this data to train a deep regression network. Our results show that we are able to recognize anatomical structure through the use of ultrasound imaging and give the user guidance toward obtaining an ideal image.

Download here

Past Publications

E.Kim, J.Yarnall, P.Shah, G.Kenyon, “A Neuromorphic Sparse Coding Defense to Adversarial Images”, International Conference on Neuromorphic Systems, ICONS, 2019.

E.Kim, E.Lawson, K.Sullivan, G.Kenyon, “Spatiotemporal Sequence Memory for Prediction using Deep Sparse Coding”, Neuro-inspired Computational Elements Workshop, NICE, 2019

J.Springer, C.Strauss, A.Thresher, E.Kim, G.Kenyon, “Classifiers Based on Deep Sparse Coding Architectures are Robust to Deep Learning Transferable Examples”, arXiv:1811.07211, 2018.

E.Kim, K.McCoy, “Multimodal Deep Learning using Images and Text for Information Graphic Classification”, ACM SIGACCESS Conference on Computers and Accessibility, Assets, 2018. (Best Paper Nominee) PDF

E.Kim, D.Hannan, G.Kenyon, “Deep Sparse Coding for Invariant Multimodal Halle Berry Neurons”, International Conference on Computer Vision and Pattern Recognition, CVPR, 2018. PDF