Satellite image analysis is revolutionizing fields like urban planning, environmental conservation, and disaster management, but achieving high-precision object detection remains a challenge. Among deep learning archi...
详细信息
Binary neural networks have become a promising research topic due to their advantages of fast inference speed and low energy consumption. However, most existing studies focus on binary convolutional neural networks, w...
详细信息
Binary neural networks have become a promising research topic due to their advantages of fast inference speed and low energy consumption. However, most existing studies focus on binary convolutional neural networks, while less attention has been paid to binary graph neural networks. A common drawback of existing studies on binary graph neural networks is that they still include lots of inefficient full-precision operations in multiplying three matrices and are therefore not efficient enough. In this paper, we propose a novel method, called re-quantization-based binary graph neural networks(RQBGN), for binarizing graph neural networks. Specifically, re-quantization, a necessary procedure contributing to the further reduction of superfluous inefficient full-precision operations, quantizes the results of multiplication between any two matrices during the process of multiplying three matrices. To address the challenges introduced by requantization, in RQBGN we first study the impact of different computation orders to find an effective one and then introduce a mixture of experts to increase the model capacity. Experiments on five benchmark datasets show that performing re-quantization in different computation orders significantly impacts the performance of binary graph neural network models, and RQBGN can outperform other baselines to achieve state-of-the-art performance.
HfO2-based ferroelectric（FE） devices have emerged as promising candidates for next-generation non-volatile memories（NVMs）, owing to their excellent CMOS compatibility, robust scalability, and low operating voltage...
HfO2-based ferroelectric（FE） devices have emerged as promising candidates for next-generation non-volatile memories（NVMs）, owing to their excellent CMOS compatibility, robust scalability, and low operating voltage requirements [1]. Among them, ultra-thin HfxZr1-xO2（HZO） FE films are particularly attractive for back-end-of-line（BEOL）integration in monolithic 3D memory architectures.
Malware detection is one of the critical tasks of cybersecurity, especially considering the growing popularity of mobile devices. The integrity and security of mobile ecosystems rely on the capacity to identify malwar...
详细信息
In computer vision,convolutional neural networks have a wide range of *** representmost of today’s data,so it’s important to know how to handle these large amounts of data *** neural networks have been shown to solv...
详细信息
In computer vision,convolutional neural networks have a wide range of *** representmost of today’s data,so it’s important to know how to handle these large amounts of data *** neural networks have been shown to solve image processing problems ***,when designing the network structure for a particular problem,you need to adjust the hyperparameters for higher *** technique is time consuming and requires a lot of work and domain *** a convolutional neural network architecture is a classic NP-hard optimization *** the other hand,different datasets require different combinations of models or hyperparameters,which can be time consuming and *** approaches have been proposed to overcome this problem,such as grid search limited to low-dimensional space and queuing by random *** address this issue,we propose an evolutionary algorithm-based approach that dynamically enhances the structure of Convolution Neural Networks(CNNs)using optimized *** study proposes a method using Non-dominated sorted genetic algorithms(NSGA)to improve the hyperparameters of the CNN *** addition,different types and parameter ranges of existing genetic algorithms are *** study was conducted with various state-of-the-art methodologies and *** have shown that our proposed approach is superior to previous methods in terms of classification accuracy,and the results are published in modern computing literature.
Internet of Medical Things (IoMT) has garnered significant research attention from both academic and medical institutions. However, the sensitive medical data involved in IoMT raises security and privacy concerns. To ...
详细信息
Offline reinforcement learning(RL) has gathered increasing attention in recent years, which seeks to learn policies from static datasets without active online exploration. However, the existing offline RL approaches o...
详细信息
Offline reinforcement learning(RL) has gathered increasing attention in recent years, which seeks to learn policies from static datasets without active online exploration. However, the existing offline RL approaches often require a large amount of pre-collected data and hence are hardly implemented by a single agent in practice. Inspired by the advancement of federated learning(FL), this paper studies federated offline reinforcement learning(FORL),whereby multiple agents collaboratively carry out offline policy learning with no need to share their raw ***, a straightforward solution is to simply retrofit the off-the-shelf offline RL methods for FL, whereas such an approach easily overfits individual datasets during local updating, leading to instability and subpar performance. To overcome this challenge, we propose a new FORL algorithm, named model-free(MF)-FORL, that exploits novel“proximal local policy evaluation” to judiciously push up action values beyond local data support, enabling agents to capture the individual information without forgetting the aggregated knowledge. Further, we introduce a model-based variant, MB-FORL, capable of improving the generalization ability and computational efficiency via utilizing a learned dynamics model. We evaluate the proposed algorithms on a suite of complex and high-dimensional offline RL benchmarks, and the results demonstrate significant performance gains over the baselines.
Energy and environmental concerns have fostered the era of electric vehicles (EVs) to take over and be welcomed more than ever. Fuel-powered vehicles are still predominant;however, this trend appears to be changing so...
详细信息
Energy and environmental concerns have fostered the era of electric vehicles (EVs) to take over and be welcomed more than ever. Fuel-powered vehicles are still predominant;however, this trend appears to be changing sooner than we might expect. Countries in Europe, Asia, and many states in America have already made the decision to transition to a fully EV industry in the next few years. This looks promising;however, drivers still have concerns about the battery mileage of such vehicles and the anxiety that such driving experiences! Indeed, driving with the probability of having insufficient battery charge that may be involved in guaranteeing the delivery to the trip destination imposes a level of anxiety on the vehicle drivers. Therefore, for an alternative to traditional fuel-powered vehicles to be convincing, there needs to be sufficient coverage of charging stations to serve cities in the same way that fuel stations serve traditional vehicles. The current navigation models select routes based solely on distance and traffic metrics, without taking into account the coverage of fuel service stations that these routes may offer. This assumption is made under the belief that all routes are adequately covered. This might be true for fuel-powered vehicles, but not for EVs. Hence, in this work, we are presenting AFARM, a routing model that enables a smart navigation system specifically designed for EVs. This model routes the EVs via paths that are lined with charging stations that align with the EV’s current charge requirements. Different from the other models proposed in the literature, AFARM is autonomous in the sense that it determines navigation paths for each vehicle based on its make, model, and current battery status. Moreover, it employs Dijkstra’s algorithm to accommodate varying least-cost navigation preferences, ranging from shortest-distance routes to routes with the shortest trip time and routes with maximum residual battery capacities as well. According to t
Lung cancer, a leading cause of mortality, necessitates prompt detection for improved patient outcomes. Recent strides in diagnostic technologies, including biomarkers, genetic testing, and advanced imaging like CT an...
详细信息
Network updates have become increasingly prevalent since the broad adoption of software-defined networks(SDNs)in data *** TCP designs,including cutting-edge TCP variants DCTCP,CUBIC,and BBR,however,are not resilient t...
详细信息
Network updates have become increasingly prevalent since the broad adoption of software-defined networks(SDNs)in data *** TCP designs,including cutting-edge TCP variants DCTCP,CUBIC,and BBR,however,are not resilient to network updates that provoke flow *** this paper,we first demonstrate that popular TCP implementations perform inadequately in the presence of frequent and inconsistent network updates,because inconsistent and frequent network updates result in out-of-order packets and packet drops induced via transitory congestion and lead to serious performance *** look into the causes and propose a network update-friendly TCP(NUFTCP),which is an extension of the DCTCP variant,as a *** are used to assess the proposed *** findings reveal that NUFTCP can more effectively manage the problems of out-of-order packets and packet drops triggered in network updates,and it outperforms DCTCP considerably.
暂无评论