Transformer has emerged as a leading architecture in neural language processing (NLP) and computer vision (CV). However, the extensive use of nonlinear operations, like softmax, poses a performance bottleneck during transformer inference and comprises up to 40% of the total latency. Hence, we propose innovations at the circuit, algorithm and architecture levels to accelerate the transformer. At the circuit level, we propose Topkima—combining top-kactivation selection with in-memory ADC (IMA) to implement efficient softmax without any sorting overhead. Only theklargest activations are sent to softmax calculation block, reducing the huge computational cost of softmax. At the algorithmic level, a modified training scheme utilizes top-kactivations only during the forward pass, combined with a sub-top-kmethod to address the crossbar size limitation by aggregating each sub-top-kvalues as global top-k. At the architecture level, we introduce a fine pipeline for efficiently scheduling data flows and an improved scale-free technique for removing scaling cost. The combined system, dubbed Topkima-Former, enhances$1.8\times -84\times $speedup and$1.2\times -36\times $energy efficiency (EE) over prior In-memory computing (IMC) accelerators. Compared to a conventional softmax macro and a digital top-k(Dtopk) softmax macro, our proposed Topkima softmax macro achieves about$15\times $and$8\times $faster speed respectively. Experimental evaluations demonstrate minimal (0.42% to 1.60%) accuracy loss for different models in both vision and NLP tasks.

Synthetic Aperture Radar (SAR) becomes prevailing in remote sensing while SAR images are challenging to interpret by human visual perception due to the active imaging mechanism and speckle noise. Recent researches on SAR-to-optical image translation provide a promising solution and have attracted increasing attentions, though still suffering from low optical image quality with geometric distortion due to the large domain gap. In this paper, we mitigate this issue from a novel perspective, i.e., neural partial differential equations (PDE). First, based on the efficient numerical scheme for solving PDE, i.e., Taylor Central Difference (TCD), we devise a basic TCD residual block to build the backbone network, which promotes the extraction of useful information in SAR images by aggregating and enhancing features from different levels. Furthermore, inspired by the Perona-Malik Diffusion (PMD), we devise a PMD neural module to implement feature diffusion through layers, aiming at removing the noises in smooth regions while preserving the geometric structures. Assembling them together, we propose a novel SAR-to-Optical image translation network named S2O-NPDE, which delivers optical images with finer structures and less noise while enjoying an explainability advantage from explicit mathematical derivation. Experiments on the popular SEN1-2 dataset show that our model outperforms state-of-the-art methods in terms of both objective metrics and visual quality.

Along with the development of social network, more and more people know the world by reading news. The problem about what kind of emotion is inspired when people read news is very worthy of discussion. This paper will mix Deep Belief Networks (DBN) model and Support Vector Machine (SVM) to a hybrid neural network model by using the Contrast Divergence (CD) algorithm to estimate the weights when training a generating model, ensure that each layer of the Restricted Boltzmann Machine (RBM) mapping the features of the inputs to the best. At the same time, we cascade the last layer of DBN and a SVM classifier to adjust judging performance. And a set of tags will be attached to the top (Associative Memory), through a process of parameter tuning, learn the identifying weights to obtain a network for the task of text classification. The experimental results show that the hybrid neural network model works better than the traditional text categorization method based on simple characteristics (such as CHI), and it is more suitable for extracting text semantic characteristics.

To improve the modeling performance — such as accuracy and robustness — of artificial neural network (ANN), a new combined ANN and corresponding optimal modeling method are proposed in this paper. The combined ANN consists of two parallel sub-networks, and methods such as "early stopping" and "data resampling" are jointly used in training process to reduce the sensitivity of the modeling performance to its structure. To achieve better performance, the structure of combined ANN is proposed to be adjusted dynamically according to the information of expectation error and real error. Simulation experimental results verify that the optimal modeling method using combined ANN can achieve much better performance than the traditional method.