This study explores 7‐methylquinolinium halobismuthates (I, Br, and Cl) with a focus on: (1) the impact of halide composition on their structural and semiconducting properties, and (2) their memristive behavior and plasticity for neuromorphic and reservoir computing. Crystallographic analysis reveals that halide substitution leads to the formation of low‐dimensional bismuth‐based frameworks. Optical bandgaps, measured via diffuse reflectance spectroscopy, correlate well with density functional theory calculations. Due to solubility constraints, only bismuth iodide complexes are integrated into electronic devices. Current–voltage measurements reveal pinched hysteresis loops, indicative of memristive behavior. Conductivity versus temperature analysis suggests both ionic and electronic conduction pathways. Given their ability to function as synaptic analogs, further tests on transient conductance (potentiation–depression) and spike‐time‐dependent plasticity are performed. The observed nonlinearity and fading memory characteristics highlight their potential for physical reservoir computing. To evaluate system‐level behavior, a device with 16 gold electrodes (1 input, 15 outputs) is fabricated on a silicon substrate coated with the target compound. The device is assessed through benchmark tasks including waveform generation, NARMA‐2, memory capacity, and noise response under DC/AC signals. Finally, the system demonstrates 82.26% accuracy in MNIST digit classification and 82% accuracy in spoken digit “2” recognition across six different speakers.
The importance of haptic in‐sensor computing devices has been increasing. Herein, a haptic sensor with a hierarchical structure is successfully fabricated via the sacrificial template method, using carbon nanotube‐polydimethylsiloxane (CNT‐PDMS) nanocomposites for in‐sensor computing applications. The CNT‐PDMS nanocomposite sensors, with different sensitivities, are obtained by varying the amount of CNTs. The input stimuli are transformed into higher‐dimensional information, enabling a new path for the CNT‐PDMS nanocomposite application, which is implemented on a robotic hand as an in‐sensor computing device by applying a reservoir computing paradigm. The nonlinear output data obtained from the sensors are trained using linear regression and used to classify nine different objects used in everyday life with an object recognition accuracy of >80% for each object. This approach can enable tactile sensation in robots while reducing the computational cost.
1 January 2024 in Nonlinear Theory and Its Applications, IEICE
Neuromorphic devices have diverse potential applications, such as brain-inspired computers and promising high-performance arithmetic systems with power saving. Reservoir computing (RC), a type of recurrent neural network (RNN), achieves learning by adjusting the weights between the intermediate and output layers. Time-delay reservoir computing introduces a delay and creates virtual nodes within the middle layer. The Ag/Ag2S nanoparticles function as nonlinear electrical devices, following the atomic switch principles of the time-delay system. Voice recognition was performed with 87.81% accuracy when six different people pronounced the same number, and 80.18% when the same person pronounced ten different numbers.
1 April 2025 in Japanese Journal of Applied Physics
Physical reservoir computing (PRC) harnesses the intrinsic nonlinear dynamics of physical systems for efficient temporal data processing, offering significant advantages in energy-efficient hardware implementation. This study explores the potential of oriented semiconducting polymer (SCP) thin films as reservoirs for PRC, focusing on two types of SCP benzo[c]cinnoline-based conjugated polymer diketopyrrolopyrrole benzo[c]cinnoline p(DPP-BZC) and regioregular poly(3-hexyl thiophene) (RR-P3HT). To enable anisotropic charge transport, uniaxially oriented thin films with edge-on molecular orientation were fabricated using the floating film transfer method. The films were electrically evaluated for anisotropic nonlinear responses, phase-shifting capabilities, and high-dimensional mapping in PRC tasks. Performance metrics, including waveform generation accuracy, were systematically investigated under varying device configurations and molecular structures. The study underscores the critical role of different conjugated polymers and their orientations in PRC performance, paving the way for developing next-generation materials for temporal signal processing and low-power intelligent hardware.
Physical reservoir computing (PRC) is an innovative computational paradigm that leverages intrinsic nonlinearity of physical systems to efficiently perform complex tasks. It is discovered that the intrinsically disordered domain structure in multiferroic YMnO 3 provides significant nonlinearity, making it a promising candidate for robust PRC with tuneability and functionality at high temperatures. This work explores the potential of YMnO 3 single crystals for PRC. PRC performance of YMnO 3 is systematically evaluated by analysing its nonlinear responses, phase shifts, and high dimensionality through benchmark tasks such as waveform generation (WG), memory capacity (MC), and second‐order nonlinear autoregressive moving average (NARMA2) time‐series prediction. This results demonstrate that YMnO 3 single crystals exhibit superior performance in these tasks, achieving high accuracy and low power consumption (≈1.77 µW and ≈0.02 nW/domain). These crystals also performed well in practical application of low‐power speech recognition. These findings establish YMnO 3 as a viable platform for next‐generation PRC technologies, addressing critical challenges in the field.