Biomembrane‐Based Memcapacitive Reservoir Computing System for Energy‐Efficient Temporal Data Processing

Abstract

Reservoir computing is a highly efficient machine learning framework for processing temporal data by extracting input features and mapping them into higher dimensional spaces. Physical reservoirs have been realized using spintronic oscillators, atomic switch networks, volatile memristors, etc. However, these devices are intrinsically energy‐dissipative due to their resistive nature, increasing their power consumption. Therefore, memcapacitive devices can provide a more energy‐efficient approach. Herein, volatile biomembrane‐based memcapacitors are leveraged as reservoirs to solve classification tasks and process time series in simulation and experimentally. This system achieves a 99.6% accuracy for spoken‐digit classification and a normalized mean square error of in a second‐order nonlinear regression task. Furthermore, to showcase the device's real‐time temporal data processing capability, a 100% accuracy for an epilepsy detection problem is achieved. Most importantly, it is demonstrated that each memcapacitor consumes an average of 41.5 fJ of energy per spike, regardless of the selected input voltage pulse width, while maintaining an average power of 415 fW for a pulse width of 100 ms, orders of magnitude lower than those achieved by state‐of‐the‐art devices. Lastly, it is believed that the biocompatible, soft nature of our memcapacitor renders it highly suitable for computing applications in biological environments.

Brain-Inspired Reservoir Computing Using Memristors with Tunable Dynamics and Short-Term Plasticity

“…It is important to note that in such times of rising demands for energy-efficient systems, novel research approaches addressed this need through either power-consumption reduction via ultralow power reservoirs 10,39 or in-device renewable energy harvesting realized in self-powered memristors. 40 Similarly, we believe that our work, which pushes for more engineering heterogeneous reservoirs, is a pivotal milestone toward realizing more power-efficient physical RC systems via the resulting minimization of overhead and footprint associated with preprocessing techniques along with the memristors' overall low power consumption.…”

Dynamic memristor for physical reservoir computing

“…In applications such as biomedical and brain–computing interfaces that require direct contact with the biological environment, memristors of biomolecules, ions, and nanofluids are attracting attention due to their chemical similarities, operating mechanisms, and operating voltages similar to those of natural synapses, such as membrane capacitors and doped lipid membranes. 47,48 Najem et al described an artificial phospholipid membrane doped with Alamethicin peptides. 49 At sufficiently low Alamethicin concentrations and subthreshold transmembrane potentials, these peptides adsorb parallel to the membrane surface without affecting their insulating conductivity states.…”

Reservoir Computing Using Interfacial Memristors with Native SiOx Nanostructures Modified by Room-Temperature Plasma Oxidation

“…In contrast, memristive devices that modulate the distribution of oxygen ions across the entire active interface tend to exhibit less variability . Recent hardware implementations of RC have harnessed the internal dynamics of interfacial memristors, showing great promise for high-efficiency memristor-based reservoir computing systems. ,,− However, these interfacial memristors have primarily relied on Schottky-like complex semiconductor interfaces, , which may not be easily compatible with scaled CMOS nodes. Metal oxide bilayer structures have also been employed for interfacial resistive switching but often require a multistep process and a relatively thick switching medium, − typically up to 220 nm.…”