Probabilistic Spike Propagation for Efficient Hardware Implementation of Spiking Neural Networks.

Spiking neural networks (SNNs) have gained considerable attention in recent years due to their ability to model temporal event streams, be trained using unsupervised learning rules, and be realized on low-power event-driven hardware. Notwithstanding the intrinsic desirable attributes of SNNs, there is a need to further optimize their computational efficiency to enable their deployment in highly resource-constrained systems. The complexity of evaluating an SNN is strongly correlated to the spiking activity in the network, and can be measured in terms of a fundamental unit of computation, viz. spike propagation along a synapse from a single source neuron to a single target neuron. We propose probabilistic spike propagation, an approach to optimize rate-coded SNNs by interpreting synaptic weights as probabilities, and utilizing these probabilities to regulate spike propagation. The approach results in 2.4-3.69× reduction in spikes propagated, leading to reduced time and energy consumption. We propose Probabilistic Spiking Neural Network Application Processor (P-SNNAP), a specialized SNN accelerator with support for probabilistic spike propagation. Our evaluations across a suite of benchmark SNNs demonstrate that probabilistic spike propagation results in 1.39-2× energy reduction with simultaneous speedups of 1.16-1.62× compared to the traditional model of SNN evaluation.

An inertial sensor-based system to develop motor capacity in children with cerebral palsy.

Learning to communicate with alternative augmentative communication devices can be difficult because of the difficulty of achieving controlled interaction while simultaneously learning to communicate. What is needed is a device that harnesses a child's natural motor capabilities and provides the means to reinforce them. We present a kinematic sensor-based system that learns a child's natural gestural capability and allows him/her to practice those capabilities in the context of a game. Movement is captured with a single kinematic sensor that can be worn anywhere on the body. A gesture recognition algorithm interactively learns gesture models using kinematic data with the help of a nearby teacher. Learned gesture models are applied in the context of a game to help the child practice gestures to gain better consistency. The system was successfully tested with a child over two sessions. The system learned four candidate gestures: lift hand, sweep right, twist right and punch forward. These were then used in a game. The child showed better consistency in performing the gestures as each session progressed. We aim to expand on this work by developing qualitative scores of movement quality and quantifying algorithm accuracy on a larger population over long periods of time.