The rapid evolution of artificial intelligence (AI) in IoT applications necessitates the execution of inference tasks on edge devices. However, the deployment of computation-intensive neural networks on resource-constrained edge systems presents a significant challenge. This brief presents the RV-WINO processor, the first silicon implementation of a RISC-V processor based on the Winograd algorithm for convolution and general matrix multiplication (GEMM) acceleration. The processor incorporates a Winograd module, which significantly reduces multiplication operations during convolutions, leading to a substantial decrease in energy consumption. In addition, the processor includes a matrix multiplication module that reuses the multipliers of the Winograd module, accelerating fully connected and dot product operations in neural networks. The RV-WINO processor fabricated in a 55-nm CMOS process achieves the peak computational performance of 0.95 and 2.39 GOPS in INT32 and INT8 modes, with its peak energy efficiency reaching 112 and 237 GOPS/W. In convolutional neural network (CNN) inference tasks, the execution time is reduced by over 80% compared with the baseline processor.

Logic Gate Networks (LGNs) exploit the similarity between neural networks and logic circuit networks and replace the neurons with logic gates.Consequently, the computation inside the neurons can be replaced by Boolean operations (16 operations for two-input logic).LGNs can be implemented by logic-based instructions in processors and significantly reduce the computation overhead during inference.However, the encoding and decoding processes at the input and output stages of LGNs face efficiency challenges when using traditional RISC-V instruction sets.This limitation arises because these processes rely on one-bit operations, which cannot fully utilize the 32-bit bandwidth of standard instructions.In this work, we proposed four custom RISC-V-based instructions to accelerate the encoding and decoding processes of LGNs.An applicationspecific RISC-V processor, called RV-LGN, has been implemented on FPGA and synthesized using Synopsys® Design Compiler with the CMOS 55nm process.The custom instructions can be called via in-line assembly in C code, making RV-LGN highly promising for implementation in edge devices.Benchmark tests on MIT-BIH, MNIST, and CIFAR-10 classification tasks demonstrate that RV-LGN achieves a runtime reduction of over 87% compared to a generic RISC-V RV32IM ISA processor.Additionally, power consumption during LGN inference is significantly reduced.For the MIT-BIH dataset, the energy consumption is 0.098 µJ/Beat, while MNIST and CIFAR-10 tasks require 0.18 µJ/Image and 0.51 µJ/Image, respectively.These results highlight the superior efficiency of RV-LGN compared to other processors.

In recommender systems, a common problem is the presence of various biases in the collected data, which deteriorates the generalization ability of the recommendation models and leads to inaccurate predictions. Doubly robust (DR) learning has been studied in many tasks in RS, with the advantage that unbiased learning can be achieved when either a single imputation or a single propensity model is accurate. In this paper, we propose a multiple robust (MR) estimator that can take the advantage of multiple candidate imputation and propensity models to achieve unbiasedness. Specifically, the MR estimator is unbiased when any of the imputation or propensity models, or a linear combination of these models is accurate. Theoretical analysis shows that the proposed MR is an enhanced version of DR when only having a single imputation and propensity model, and has a smaller bias. Inspired by the generalization error bound of MR, we further propose a novel multiple robust learning approach with stabilization. We conduct extensive experiments on real-world and semi-synthetic datasets, which demonstrates the superiority of the proposed approach over state-of-the-art methods.

Click-through rate (CTR) prediction is one of the fundamental tasks in online advertising and recommendation. Multi-layer perceptron (MLP) serves as a core component in many deep CTR prediction models, but it has been widely shown that applying a vanilla MLP network alone is ineffective in learning complex feature interactions. As such, many two-stream models (e.g., Wide&Deep, DeepFM, and DCN) have recently been proposed, aiming to integrate two parallel sub-networks to learn feature interactions from two different views for enhanced CTR prediction. In addition to one MLP stream that learns feature interactions implicitly, most of the existing research focuses on designing another stream to complement the MLP stream with explicitly enhanced feature interactions. Instead, this paper presents a simple two-stream feature interaction model, namely FinalMLP, which employs only MLPs in both streams yet achieves surprisingly strong performance. In contrast to sophisticated network design in each stream, our work enhances CTR modeling through a feature selection module, which produces differentiated feature inputs to two streams, and a group-wise bilinear fusion module, which effectively captures stream-level interactions across two streams. We show that FinalMLP achieves competitive or even better performance against many existing two-stream CTR models on four open benchmark datasets and also brings significant CTR improvements during an online A/B test in our industrial news recommender system. We envision that the simple yet effective FinalMLP model could serve as a new strong baseline for future development of two-stream CTR models. Our source code will be available at MindSpore/models and FuxiCTR/model_zoo.

In this paper, we study a cooperative perception scheme in a vehicular sensor network, attempting to fuse the semantic information provided by different sensors at multiple vehicles, so as to expand the vehicle’s perception range, eliminate blind spots, improve the ability to handle environmental interference and enhance the accuracy and robustness of the perception results. The key to guide the fusion process is the evaluation of the confidence levels of the outputs provided by various machine learning schemes implemented at individual sensors in the vehicular sensor network. We first propose an evaluation criterion termed as Environmental Sensitivity (ES), which is used to measure the sensitivity of the network to environmental changes. Based on the ES, we further evaluate the confidence of the perception output of neural networks and quantify the confidence level considering the abnormal level of the input data, the general performance of the perception algorithm, the detection performance and the ES of the network. Semantic information fusion algorithm is then developed based upon the confidence levels. Experiment results are provided to validate the proposed fusion method in various scenarios.

In autonomous driving, environment perception is the fundamental task for intelligent vehicles which provides the necessary environment information for other applications. The main issues in existing environment perception can be categorized into two aspects. On the one hand, all sensors are prone to measurement errors and failures. On the other hand, in complex driving environments, vehicles may encounter a variety of blind spots caused by vehicle occlusions, overlaps, and harsh weather conditions, which will cause sensors to experience low-quality data or to miss crucial environmental information. To cope with these issues, a multivehicle and multisensor (MVMS) cooperative perception method is presented to construct the occupancy grid map (OGM) of vehicles in a global view for the environment perception of autonomous driving. Distinct from existing environment perception methods, our proposed MVMS-OGM not only provides continuous geographical information but also captures and fuses continuous information with soft occupancy probabilities, resulting in more comprehensive and raw environmental information. Simulations and real-world experiments demonstrate that the proposed approach not only expands the perception range in comparison with single-vehicle sensing but also better captures the uncertainty of sensor data by fusing the occupancy probabilities with soft information.

For autonomous driving of unmanned vehicles in intelligent transportation systems, multi-vehicle cooperative perception supported by vehicular networks can greatly improve the accuracy and reliability of the perception decisions. Currently, the perception decisions for a single vehicle are mostly provided by neural networks. Therefore, in order to fuse the perception decisions from multiple vehicles, the credibility of the neural network outputs needs to be studied. Among various factors, the environment is one of the most important affecting vehicles' perception decisions. In this paper, we propose a new evaluation criteria for the neural networks used in the perception module of unmanned vehicles. This criterion is termed as Environmental Sensitivity (ES), indicates the sensitivity of the network to environmental changes. We design an algorithm to quantitatively measure the ES value of different perception networks based on the extracted features. Experimental results show that our algorithm can well capture the sensitivity of the network in different environments and the ES values will be helpful to the subsequent decision fusion process.