3D Network on Chips (3D-NoCs) have ample on-chip wiring resources and high bandwidth, yet face numerous hotspots and higher temperature gradients due to increased integration and power density. This could lead to device failure, impacting system stability. Our paper introduces a thermal- and traffic-aware mapping method for 3D-NoC-based neural network accelerators. Firstly, based on the average load of different neural network layer, we determine their mapping sequences and suitable dies. Secondly, to minimize delay and alleviate hotspot temperatures, we allocate groups to appropriate nodes. Compared with previous works, TTNNM reduces the average temperature by 3.0°C, 2.2°C, 2.4°C, temperature variance by 58.4%, 64.8%, 73.0%, maximum temperature by 9.3°C, 7.9°C, 12.0°C, and packet latency by 31.7%, 17.2%, 25.1%.

Choices of dataflows, which are known as intra-core neural network (NN) computation loop nest scheduling and inter-core hardware mapping strategies, play a critical role in the performance and energy efficiency of NoC-based neural network accelerators. Confronted with an enormous dataflow exploration space, this paper proposes an automatic framework for generating and optimizing the full-layer-mappings based on two reinforcement learning algorithms including A2C and PPO. Combining soft and hard constraints, this work transforms the mapping configuration into a sequential decision problem and aims to explore the performance and energy efficient hardware mapping for NoC systems. We evaluate the performance of the proposed framework on 10 experimental neural networks. The results show that compared with the direct-X mapping, the direct-Y mapping, GA-base mapping, and NN-aware mapping, our optimization framework reduces the average execution time of 10 experimental NNs by 9.09$\%$, improves the throughput by 11.27$\%$, reduces the energy by 12.62$\%$, and reduces the time-energy-product (TEP) by 14.49$\%$. The results also show that the performance enhancement is related to the coefficient of variation of the neural network to be computed.

Text-based person search aims to retrieve images of a certain pedestrian by a textual description. The key challenge of this task is to eliminate the inter-modality gap and achieve the feature alignment across modalities. In this paper, we propose a semantic-aligned embedding method for text-based person search, in which the feature alignment across modalities is achieved by automatically learning the semantic-aligned visual features and textual features. First, we introduce two Transformer-based backbones to encode robust feature representations of the images and texts. Second, we design a semantic-aligned feature aggregation network to adaptively select and aggregate features with the same semantics into part-aware features, which is achieved by a multi-head attention module constrained by a cross-modality part alignment loss and a diversity loss. Experimental results on the CUHK-PEDES and Flickr30K datasets show that our method achieves state-of-the-art performances.

In the past few years, cross-modal image-text retrieval (ITR) has experienced increased interest in the research community due to its excellent research value and broad real-world application. It is designed for the scenarios where the queries are from one modality and the retrieval galleries from another modality. This paper presents a comprehensive and up-to-date survey on the ITR approaches from four perspectives. By dissecting an ITR system into two processes: feature extraction and feature alignment, we summarize the recent advance of the ITR approaches from these two perspectives. On top of this, the efficiency-focused study on the ITR system is introduced as the third perspective. To keep pace with the times, we also provide a pioneering overview of the cross-modal pre-training ITR approaches as the fourth perspective. Finally, we outline the common benchmark datasets and evaluation metric for ITR, and conduct the accuracy comparison among the representative ITR approaches. Some critical yet less studied issues are discussed at the end of the paper.

Unsynchronization of the separate transmit–receive beams makes it a challenging task to obtain high-quality images for nonparallel-track bistatic synthetic aperture radar (NP-BiSAR). To accommodate this issue, we propose an imaging configuration where the receiver’s beam follows the transmitter’s one actively by adjusting the squint angle of receiver. And a frequency extended nonlinear chirp scaling (FENLCS) algorithm is modified to cope with new effects introduced by the innovative configuration, which is based on an improved quadratic ellipse model. Based on the new model, some innovative improvements on image formation are made, including a residual azimuth-dependent high-order range cell migration correction (ADH-RCMC) and a rederived FENLCS that takes highly varying Doppler centroid into consideration, which contribute to better imaging quality. Simulation results validate the effectiveness of the proposed configuration and algorithm.

Small playback lag in live streaming is important for time-critical and interactive applications such as live stock, market updates, sports and remote education. In this paper, we present Elite addresses the playback lag problem in peer-assisted live streaming systems. Instead of deploying a large initial offset to all the users, Elite seeks the possibility of initializing users with layered proportional initial scheduling point, thus achieving differentiated playback lag service for the system. For saving server bandwidth and reducing lag time, Elite employs a novel strategy which arranges peers into a virtual tree structure and quantifies playback lag of each layer that finally converges to a constant value. This way, Elite can help users to achieve much shorter average playback lag and prioritized service within the same channel. As illustrated in our design, analysis, and simulation studies, Elite is able to fully exploit limited pool of server bandwidth to support peer-assisted live streaming with prioritized playback lag, and achieves shorter average playback lag compared with synchronized strategy, such as R2.

H.264 is an advanced video compression standard, absorbing the advantages of the previous standards. In this paper, block-based stereoscopic video coding is studied, and some schemes of using H.264 are discussed. The stereoscopic video coding methods based on H.264 and based on H.263+ are also compared by the experiments, experimental results shove that the former is more effective than the latter in compression efficiency and image quality, and the H.264-based stereoscopic video coding scheme with temporal scalability is quite effective.

The author is concerned with a class of nonstationary processes described by the piecewise ARMA (autoregressive moving-average) models, whose parameters change abruptly (or jump) at some unknown times. Two different models are examined and the statistics of this type of processes are studied. It is shown that the means of such processes do not depend on the jumps, while the autocovariance functions change gradually after the parameter jump and follow certain interesting patterns.>

A method is proposed to improve the accuracy of the parameter estimates of highly damped sinusoidal signals in additive noise. The method first utilizes singular value decomposition (SVD) of the signal matrix to reduce the noise effect, then based on the noise-reduced signals the weighted sum of squared errors is minimized using an adaptive lattice filter algorithm to estimate the parameters in the backward linear prediction model. In the case of large damping factors, the experimental results show that the proposed method gives more accurate estimates than the eigenvector method.>

We define a piecewise AR model For a class of time series whose statistical properties change abruptly at some unknown time points. For such a model we consider the problems of jump detection and fast tracking of the changing parameters. A method based on the adaptive least squares lattice filter algorithm is proposed. The method automatically detects the occurrences of jumps and adjusts the adaptation rate of the adaptive lattice algorithm, so both accurate estimates and fast tracking ability are achieved.