Real-world sensory-processing applications require compact, low-latency, and low-power computing systems. Enabled by their in-memory event-driven computing abilities, hybrid memristive-Complementary Metal-Oxide Semiconductor neuromorphic architectures provide an ideal hardware substrate for such tasks. To demonstrate the full potential of such systems, we propose and experimentally demonstrate an end-to-end sensory processing solution for a real-world object localization application. Drawing inspiration from the barn owl’s neuroanatomy, we developed a bio-inspired, event-driven object localization system that couples state-of-the-art piezoelectric micromachined ultrasound transducer sensors to a neuromorphic resistive memories-based computational map. We present measurement results from the fabricated system comprising resistive memories-based coincidence detectors, delay line circuits, and a full-custom ultrasound sensor. We use these experimental results to calibrate our system-level simulations. These simulations are then used to estimate the angular resolution and energy efficiency of the object localization model. The results reveal the potential of our approach, evaluated in orders of magnitude greater energy efficiency than a microcontroller performing the same task.
The connectivity in the brain is locally dense and globally sparse - giving rise to a small-world graph. This is a principle that has persisted during the evolution of many species - indicating a universal solution to the efficient routing of information. However, existing circuit architectures for artificial neural networks neither leverage this organization nor do they efficiently support small-world neural network models. Here, we propose the neuromorphic Mosaic: a non-von Neumann systolic architecture that uses distributed memristors, not only for in-memory computing, but also for in-memory routing, to efficiently implement small-world graph topologies. We design, fabricate, and experimentally demonstrate the building blocks of this architecture, using integrated memristors with 130 nm CMOS technology. We demonstrate that neural networks implemented following this approach can achieve competitive accuracy figures compared to equivalent unconstrained and full-precision networks, for three real-time benchmarks: classification of electrocardiography signals, keyword spotting and motor control via reinforcement learning. The Mosaic shows improvements between one and four orders of magnitude, compared to other event-based neuromorphic architectures for routing events across the network. The Mosaic opens up a new scalable approach for designing edge AI systems based on distributed computing and in-memory routing, offering a natural platform onto which architectures inspired by biological nervous systems can be readily mapped.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.