No abstract

In this paper, we propose an online learning algorithm for supervised learning in multilayer spiking neural networks (SNNs). It is found that the spike timings of neurons in an SNN can be exploited to estimate the gradients that are associated with each synapse. With the proposed method of estimating gradients, learning similar to the stochastic gradient descent process employed in a conventional artificial neural network (ANN) can be achieved. In addition to the conventional layer-by-layer backpropagation, a one-pass direct backpropagation is possible using the proposed learning algorithm. Two neural networks, with one and two hidden layers, are employed as examples to demonstrate the effectiveness of the proposed learning algorithms. Several techniques for more effective learning are discussed, including utilizing a random refractory period to avoid saturation of spikes, employing a quantization noise injection technique and pseudorandom initial conditions to decorrelate spike timings, in addition to leveraging the progressive precision in an SNN to reduce the inference latency and energy. Extensive parametric simulations are conducted to examine the aforementioned techniques. The learning algorithm is developed with the considerations of ease of hardware implementation and relative compatibility with the classic ANN-based learning. Therefore, the proposed algorithm not only enjoys the high energy efficiency and good scalability of an SNN in its specialized hardware but also benefits from the well-developed theory and techniques of conventional ANN-based learning. The Modified National Institute of Standards and Technology database benchmark test is conducted to verify the newly proposed learning algorithm. Classification correct rates of 97.2% and 97.8% are achieved for the one-hidden-layer and two-hidden-layer neural networks, respectively. Moreover, a brief discussion of the hardware implementations is presented for two mainstream architectures.

Abstract-A novel compact branch-line coupler operating in two arbitrary frequencies is proposed, analyzed and designed. Steppedimpedance stubs are used in the branch-line coupler to achieve dualband applications. Parameters of the structure are chosen and provided for design guidelines. Broader operating frequency ratios and compactness are achievable. For the purpose of validation, a microstrip coupler operating at 2.4/5.2 GHz is fabricated and measured.

The digital image segmentation algorithm based on deep learning plays an important role in the monitoring of seabed mineral resources. The traditional segmentation algorithm has insufficient performance in the face of adhesion, and the segmentation boundary is fuzzy. For this reason, an improved segmentation algorithm by learning a deep convolution network is proposed. A typical encoder-decoder structure is used to construct the network model, and the decoder part is up-sampled at different scales to obtain the final segmentation map. The performance of the algorithm is tested on the gray scale electron microscopy (EM) image dataset and the seabed mineral image dataset. The experimental shows that the Rand theoretic score can achieve 0.916 on EM image dataset, and a better segmentation result on the seabed mineral image dataset than the original U-net Convolutional Network. INDEX TERMS Image segmentation, deep learning, convolutional networks, EM images, seabed mineral images.

No abstract