SSTDP: Supervised Spike Timing Dependent Plasticity for Efficient Spiking Neural Network Training

Spiking Neural Networks (SNNs) are a pathway that could potentially empower low-power event-driven neuromorphic hardware due to their spatio-temporal information processing capability and high biological plausibility. Although SNNs are currently more efficient than artificial neural networks (ANNs), they are not as accurate as ANNs. Error backpropagation is the most common method for directly training neural networks, promoting the prosperity of ANNs in various deep learning fields. However, since the signals transmitted in the SNN are non-differentiable discrete binary spike events, the activation function in the form of spikes presents difficulties for the gradient-based optimization algorithms to be directly applied in SNNs, leading to a performance gap (i.e., accuracy and latency) between SNNs and ANNs. This paper introduces a new learning algorithm, called SSTDP, which bridges the gap between backpropagation (BP)-based learning and spike-time-dependent plasticity (STDP)-based learning to train SNNs efficiently. The scheme incorporates the global optimization process from BP and the efficient weight update derived from STDP. It not only avoids the non-differentiable derivation in the BP process but also utilizes the local feature extraction property of STDP. Consequently, our method can lower the possibility of vanishing spikes in BP training and reduce the number of time steps to reduce network latency. In SSTDP, we employ temporal-based coding and use Integrate-and-Fire (IF) neuron as the neuron model to provide considerable computational benefits. Our experiments show the effectiveness of the proposed SSTDP learning algorithm on the SNN by achieving the best classification accuracy 99.3% on the Caltech 101 dataset, 98.1% on the MNIST dataset, and 91.3% on the CIFAR-10 dataset compared to other SNNs trained with other learning methods. It also surpasses the best inference accuracy of the directly trained SNN with 25~32× less inference latency. Moreover, we analyze event-based computations to demonstrate the efficacy of the SNN for inference operation in the spiking domain, and SSTDP methods can achieve 1.3~37.7× fewer addition operations per inference. The code is available at: https://github.com/MXHX7199/SNN-SSTDP.

Metallothermic Reduction at Low Temperature Synthesis of Ti4O7 Powder with Lithium Ion Insertion/Extraction Property

In this work, a thermal reduction process via ultrafine titanium powder as the reducing agent under argon atmosphere is firstly used to prepare Ti4O7. Compared with the conventional method, this experiment process reduces the sintering temperature to 850°C. The phase transformation and the morphology of the as-prepared powders are examined by X-Ray diffraction (XRD) and scanning electron microscopy (SEM). Besides, it is found that the Ti4O7 powders obtained by titanium thermal reduction method exhibited the crystal structure, distinctly possessing an average particle size around 750 nm. The as-prepared Ti4O7 nanoparticles are used as anode active material in lithium battery. The results demonstrate that the anode with Ti4O7 calcined at 850°C by titanium thermal reduction method exhibited insertion/extraction lithium ion property.

Computer Simulation of PMSM Motor with Five Phase Inverter Control using Signal Processing Techniques

The signal processing techniques and computer simulation play an important role in the fault diagnosis and tolerance of all types of machines in the first step of design. Permanent magnet synchronous motor (PMSM) and five phase inverter with sine wave pulse width modulation (SPWM) strategy is developed. The PMSM speed is controlled by vector control. In this work, a fault tolerant control (FTC) system in the PMSM using wavelet switching is introduced. The feature extraction property of wavelet analysis used the error as obtained by the wavelet de-noised signal as input to the mechanism unit to decide the healthy system. The diagnosis algorithm, which depends on both wavelet and vector control to generate PWM as current based manage any parameter variation. An open-end phase PMSM has a larger range of speed regulation than normal PMSM. Simulation results confirm the validity and effectiveness of the switching strategy.

Effects of Nano-Particles on Resistance of DOP Migration from Flexible PVC

The effects of nano-particles (CaCO3, SiO2 and organic modified montmorillonite (OMMT) on the suppression of DOP migration from PVC matrix and mechanical properties of PVC composites were evaluated. The results indicated that the three kinds of nano-particles could improve the ability of anti-migration of DOP in flexible PVC. A certain content of nano-particle could decrease the migration rate of DOP. The addition of 5 phr SiO2 decreased the extraction rate of DOP to 15.6%, and SiO2-5/PVC composite film possessed superior anti-extraction property. The introduction of 5 phr OMMT reduced the volatilization of DOP to 0.067%, and OMMT-5/PVC exhibited the lowest volatilization of DOP. Inorganic nano-particles with high surface energy were easy to aggregate, and the influences of aggregation on the properties of composites were clearly detrimental, which results in a drastic decrease of polymer anti-migration and mechanical performance.