RESUMO
Neuro-symbolic artificial intelligence has garnered considerable attention amid increasing industry demands for high-performance neural networks that are interpretable and adaptable to previously unknown problem domains with minimal reconfiguration. However, implementing neuro-symbolic hardware is challenging due to the complexity in symbolic knowledge representation and calculation. We experimentally demonstrated a memristor-based neuro-fuzzy hardware based on TiN/TaOx/HfOx/TiN chips that is superior to its silicon-based counterpart in terms of throughput and energy efficiency by using array topological structure for knowledge representation and physical laws for computing. Intrinsic memristor variability is fully exploited to increase robustness in knowledge representation. A hybrid in situ training strategy is proposed for error minimizing in training. The hardware adapts easier to a previously unknown environment, achieving ~6.6 times faster convergence and ~6 times lower error than deep learning. The hardware energy efficiency is over two orders of magnitude greater than field-programmable gate arrays. This research greatly extends the capability of memristor-based neuromorphic computing systems in artificial intelligence.
RESUMO
To give people more specific information on the quality of their daily motion, it is necessary to continuously measure muscular activity during everyday occupations in an easy way. The traditional methods to measure muscle activity using a combination of surface electromyography (sEMG) sensors and optical motion capture system are expensive and not suitable for non-technical users and unstructured environment. For this reason, in our group we are researching methods to estimate leg muscle activity using non-contact wearable sensors, improving ease of movement and system usability. In a previous study, we developed a method to estimate muscle activity via only a single inertial measurement unit (IMU) on the shank. In this study, we describe a method to estimate muscle activity during walking via two IMU sensors, using an original sensing system and specifically developed estimation algorithms based on ANN techniques. The muscle activity estimation results, estimated by the proposed algorithm after optimization, showed a relatively high estimation accuracy with a correlation efficient of R2 = 0.48 and a standard deviation STD = 0.10, with a total system average delay of 192 ms. As the average interval between different gait phases in human gait is 250-1000 ms, a 192 ms delay is still acceptable for daily walking requirements. For this reason, compared with the previous study, the newly proposed system presents a higher accuracy and is better suitable for real-time leg muscle activity estimation during walking.
