To deeply excavate the information contained in user data and better alleviate cold start problem, we propose KGCN++, a user information enhanced knowledge graph convolutional networks model for recommender system, wh...
详细信息
Logic Synthesis (LS) plays a vital role in chip design. A key task in LS is to simplify circuits-modeled by directed acyclic graphs (DAGs)-with functionality-equivalent transformations. To tackle this task, many LS he...
详细信息
Logic Synthesis (LS) plays a vital role in chip design. A key task in LS is to simplify circuits-modeled by directed acyclic graphs (DAGs)-with functionality-equivalent transformations. To tackle this task, many LS heuristics apply transformations to subgraphs-rooted at each node on an input DAG-sequentially. However, we found that a large number of transformations are ineffective, which makes applying these heuristics highly time-consuming. In particular, we notice that the runtime of the Resub and Mfs2 heuristics often dominates the overall runtime of LS optimization processes. To address this challenge, we propose a novel data-driven LS heuristic paradigm, namely PruneX, to reduce ineffective transformations. The major challenge of developing PruneX is to learn models that well generalize to unseen circuits, i.e., the out-of-distribution (OOD) generalization problem. Thus, the major technical contribution of PruneX is the novel circuit domain generalization framework, which learns domain-invariant representations based on the transformation-invariant domain-knowledge. To the best of our knowledge, PruneX is the first approach to tackle the OOD problem in LS heuristics. We integrate PruneX with the aforementioned Resub and Mfs2 heuristics. Experiments demonstrate that PruneX significantly improves their efficiency while keeping comparable optimization performance on industrial and very large-scale circuits, achieving up to 3.1× faster runtime. Copyright 2024 by the author(s)
Federal Learning (FL) has made remarkable achievements in the medical field, but the technology still faces challenges in data sharing and privacy protection, which limits its application in the medical field. To solv...
详细信息
ISBN:
(数字)9798350377613
ISBN:
(纸本)9798350377620
Federal Learning (FL) has made remarkable achievements in the medical field, but the technology still faces challenges in data sharing and privacy protection, which limits its application in the medical field. To solve these problems, we developed the CQUPT-HDS system. The system is dedicated to solving the security and privacy protection problems in cross-domain collaborative analysis, ensuring the security of data and models. In view of the sensitivity of medical data, complexity of cross-domain sharing and privacy protection, key technologies such as local differential privacy algorithm and searchable encryption algorithm are adopted to optimize model training, data transmission and role rights management, so as to achieve secure and efficient cross-domain data sharing and collaborative computing. The purpose of introducing the system is to realize comprehensive security situation awareness and detection, build a multi-level defense system, solve the challenges of data sharing and privacy protection in the medical field, and provide a reliable solution for cross-domain sharing and collaborative computing of medical data.
Graph neural networks (GNNs) are specially designed to process graph data because of their ability to effectively capture complex structures and relationships within graphs. However, due to the pixel-based nature of i...
Graph neural networks (GNNs) are specially designed to process graph data because of their ability to effectively capture complex structures and relationships within graphs. However, due to the pixel-based nature of images, representing them as graphs and then using graph networks poses challenges. Inspired by granular computing, we propose a method to compute multi-granularity structured features from image tiles, which facilitates efficient and direct graph-based representation learning. Specifically: (1) We design a method to reorganize all regional features into multiple tiles and construct a multi-granularity graph representation, which includes nodes and edges. (2) We use gray-level co-occurrence matrices to extract the image texture features to compensate for the information lost during graph representation. (3) We design a graph neural network classifier that combines the above graph features and image texture features. Experimental evaluations on two benchmark image classification datasets demonstrate the efficiency of our proposed method and its potential to establish a new paradigm for image processing. Codes are available at https://***/ddw2AIGROUP2CQUPT/MGRIR.
In this paper,we first give characterizations of weighted Besov spaces with variable exponents via Peetre’s maximal *** we obtain decomposition characterizations of these spaces by atom,molecule and *** an applicatio...
详细信息
In this paper,we first give characterizations of weighted Besov spaces with variable exponents via Peetre’s maximal *** we obtain decomposition characterizations of these spaces by atom,molecule and *** an application,we obtain the boundedness of the pseudo-differential operators on these spaces.
Federated Learning (FL) has progressed, providing a distributed mechanism where data need not be consolidated, thereby enhancing the privacy and security of sensitive healthcare data. Recent advancements in multimodal...
详细信息
Multi-band optical networks are a potential technology for increasing network ***,the strong interference and non-uniformity between wavelengths in multi-band optical networks have become a bottleneck restricting the ...
详细信息
Multi-band optical networks are a potential technology for increasing network ***,the strong interference and non-uniformity between wavelengths in multi-band optical networks have become a bottleneck restricting the transmission capacity of multi-band optical *** overcome these challenges,it is particularly important to implement optical power optimization targeting wavelength ***,based on the generalized Gaussian noise model,we first formulate an optimization model for the problems of routing,modulation format,wavelength,and power allocation in C+L+S multi-band optical *** objective function is to maximize the average link capacity of the network while ensuring that the Optical Signal-to-Noise(OSNR)threshold of the service request is not ***,we propose a NonLinear Interferenceaware(NLI-aware)routing,modulation format,wavelength,and power allocation ***,we conduct simulations under different test *** simulation results indicate that our algorithm can effectively reduce the blocking probability by 23.5%and improve the average link capacity by 3.78%in C+L+S multi-band optical networks.
It is challenging to cluster multi-view data in which the clusters have overlapping *** multi-view clustering methods often misclassify the indistinguishable objects in overlapping areas by forcing them into single cl...
详细信息
It is challenging to cluster multi-view data in which the clusters have overlapping *** multi-view clustering methods often misclassify the indistinguishable objects in overlapping areas by forcing them into single clusters,increasing clustering *** solution,the multi-view dynamic kernelized evidential clustering method(MvDKE),addresses this by assigning these objects to meta-clusters,a union of several related singleton clusters,effectively capturing the local imprecision in overlapping *** offers two main advantages:firstly,it significantly reduces computational complexity through a dynamic framework for evidential clustering,and secondly,it adeptly handles non-spherical data using kernel techniques within its objective *** on various datasets confirm MvDKE's superior ability to accurately characterize the local imprecision in multi-view non-spherical data,achieving better efficiency and outperforming existing methods in overall performance.
The information conveyed through facial expressions accounts for a large proportion of the total information and can effectively express people's intentions and emotions. Facial expression recognition has laid the...
详细信息
USB interfaces have become ubiquitous in various Internet of Things (IoT) devices, all adhering to the same USB protocol. While enhancing convenience, they also widen the potential attack surface. Fuzzing is a proacti...
详细信息
暂无评论