Limited attention has been given to unsupervised capsule networks (CapsNets) with contrastive learning due to the challenge of harmoniously learning interpretable primary and high-level capsules. To address this issue...
详细信息
Limited attention has been given to unsupervised capsule networks (CapsNets) with contrastive learning due to the challenge of harmoniously learning interpretable primary and high-level capsules. To address this issue, we focus on three aspects: loss function, routing algorithm, and training strategy. First, we propose a comprehensive contrastive loss to ensure consistency in learning both high-level and primary capsules across different objects. Next, we introduce an agreement-based routing mechanism for the activation of high-level capsules. Finally, we present a two-stage training strategy to resolve conflicts between multiple losses. Ablation experiments show that these methods all improve model performance. Results from linear evaluation and semi-supervised learning demonstrate that our model outperforms other CapsNets and convolutional neural networks in learning high-level capsules. Additionally, visualizing capsules provides insights into the primary capsules, which remain consistent across images and align with human vision.
In this paper, we study the properties of the bus-based hypercube, denoted as U(n, b), which is a kind of multiple-bus networks (MBN). U(n, b) consists of 2(n) processors and 2(b) buses, where 0 less than or equal to ...
详细信息
In this paper, we study the properties of the bus-based hypercube, denoted as U(n, b), which is a kind of multiple-bus networks (MBN). U(n, b) consists of 2(n) processors and 2(b) buses, where 0 less than or equal to b less than or equal to n - 1, and each processor is connected to either inverted right perpendicularb+2/2inverted left perpendicular or inverted right perpendicularb+1/2inverted left perpendicular buses. We show that the diameter of U(n, b) is inverted right perpendicularb+1/2inverted left perpendicular if b greater than or equal to 2. We also present an algorithm to select the best neighbor processor via which we can obtain one shortest routing path. In U(n, b), we show that if there exist some faults, the fault diameter DF(n, b, f) less than or equal to b + 1, where f is the sum of bus faults and processor faults and 0 less than or equal to f less than or equal to inverted right perpendicularb-3/2inverted left perpendicular. Furthermore, we also show that the bus-fault diameter DB(n, b, f) less than or equal to b + 1, where f is the sum of bus faults and processor faults and 0 less than or equal to f less than or equal to inverted right perpendicularb-1/2inverted left perpendicular and f is the number of bus faults. These results improve significantly the previous result that DB(n, b, f) less than or equal to b + 2f + 1, where f is the number of bus faults.
Editor's notes: This article proposes a reinforcement-learning-based methodology to improve energy-efficiency of Network-on-Chip-based systems. -Partha Pratim Pande, Washington State University, USA
Editor's notes: This article proposes a reinforcement-learning-based methodology to improve energy-efficiency of Network-on-Chip-based systems. -Partha Pratim Pande, Washington State University, USA
mymargin2 With the end of Moore's law and Dennard scaling, waferscale systems or processors that integrate multiple pre-tested known good dies (KGDs) on a waferscale-interposer are new approaches to further improv...
详细信息
mymargin2 With the end of Moore's law and Dennard scaling, waferscale systems or processors that integrate multiple pre-tested known good dies (KGDs) on a waferscale-interposer are new approaches to further improve the chiplet-based system's performance. This article explores the network on wafer (NoW) architecture of waferscale switching system under several physical constraints. A software-based approach is proposed to redefine the topological property. A five-level butterfly fat-tree (BFT)-like logical topology with 8.96-Tb/s (896 ports $\times$ 10 Gb/s/port) switching bandwidth is achieved based on 2-D-mesh-like physical topology. We show that the proposed BFT-like topology with breadth-first-search (BFS) based traffic balanced routing algorithm reduces 55.6% hops, 41.4% transmission delay, and improves 24.2% throughput compared to 2-D-mesh-like topology under different traffic distributions. This BFT-like waferscale switching system is suitable for high-performance computing and data centers. In addition, the numerical analysis shows that the waferscale package can provide significant power efficiency and latency advantages compared to the typical single-chip package, which mainly benefits from the short-reach IO requirements. Note that the proposed waferscale switching system is compatible with high-switch-capacity dies with advanced process technology, which can further improve system performance. Finally, we present the physical implementations for the waferscale system with heterogeneous dies.
For a natural number n, let S-n denote the symmetric group on n letters. Let SEPn be the Cayley graph Cay(S-n, {sigma, sigma(-1), tau}), where tau = (12) and sigma = (1,..., n). In this note, we present an easy static...
详细信息
For a natural number n, let S-n denote the symmetric group on n letters. Let SEPn be the Cayley graph Cay(S-n, {sigma, sigma(-1), tau}), where tau = (12) and sigma = (1,..., n). In this note, we present an easy static routing algorithm for SEPn which gives us an exact formula for the path between every pair of nodes in SEPn and it does not need any calculations. Also, we show that an upper bound of this algorithm is l(3n + 1)(2)/9. Furthermore, we use our results to present a dynamic routing algorithm for SEPn which needs at most n calculations for routing and also it does not need large amount of memory for routing tables.
Recently, a series of researches have been emphasized on developing advanced satellite networks, mostly because of its advantage in providing spaced-based global communication service. But most of these work prefer to...
详细信息
ISBN:
(数字)9781510628274
ISBN:
(纸本)9781510628274
Recently, a series of researches have been emphasized on developing advanced satellite networks, mostly because of its advantage in providing spaced-based global communication service. But most of these work prefer to focus on the time-varying topologies, large delays and intermittent connections of satellite networks. However, there is another issue worthy of attentions, i.e., the scarcity and preciousness of satellite resources, owing to the shortage of orbit resources and the high cost of launching a satellite. Therefore, it is significantly important to consider the efficient utilization of resources during designing routing strategies for satellite networks. In this paper, we propose two routing algorithms to optimize the number of used inter-satellite links, which will directly improve the bandwidth utilization and save resources for LEO satellite networks. The basic idea is to reduce the number of links used by lower-priority traffic through scheduling them to links used by highest-priority services, and simultaneously introduce the load balancing strategies to control the aggregation of network flow. Simulation results show that with the price of little longer latency and load unbalancing, our algorithms can effectively decrease the total number of used links, and thus improve the resource utilization and save energy for satellite networks.
In this paper we describe a deterministic algorithm for solving any 1-1 packet-routing problem on an n x n mesh in 2n - 2 steps using constant-size queues. The time bound is optimal in the worst case. The best previou...
详细信息
In this paper we describe a deterministic algorithm for solving any 1-1 packet-routing problem on an n x n mesh in 2n - 2 steps using constant-size queues. The time bound is optimal in the worst case. The best previous deterministic algorithm for this problem required time 2n + Theta(n/q) using queues of size Theta(q) for any 1 less than or equal to q less than or equal to n, and the best previous randomized algorithm required time 2n + Theta(log n) using constant-size queues.
Optical network-on-chip (ONoC) is an emerging architecture to realize high-performance and low power consumption for many-core processors. Multicast in ONoC is a vital communication pattern widely used in parallel com...
详细信息
Optical network-on-chip (ONoC) is an emerging architecture to realize high-performance and low power consumption for many-core processors. Multicast in ONoC is a vital communication pattern widely used in parallel computing and computational genomics. This paper proposes an efficient scheme for multiple multicast requests in ONoC. Firstly, we design an algorithm to construct two trees for two multicasts without duplicate edges in a mesh topology, and it only uses one wavelength. We propose a graph to represent the relationship between each multicast. Secondly, a group-based wavelength assignment scheme is designed to assign wavelengths for multiple multicasts. Finally, we used data analysis to extract helpful simulation results to form conclusions. The results indicate that our algorithm is superior to other tree-based schemes in communication latency and wavelength number usage.
With rapid development of wireless communication, sensor, micro power system and electronic technology, the research on wireless sensor network has attracted more and more attention. The work proposed routing algorith...
详细信息
With rapid development of wireless communication, sensor, micro power system and electronic technology, the research on wireless sensor network has attracted more and more attention. The work proposed routing algorithm in wireless sensor network based on ant colony optimization by analyzing routing protocol and utilizing advanced idea. Ant colony optimization algorithm has advantages in implementing local work, supporting multiple paths and integrating link quality into pheromone formation. In routing selection, the work calculated probability that node is selected as the next hop according to pheromone concentration on the path. With characteristics including self-organization, dynamic and multipath, ant colony optimization algorithm is suitable for routing in wireless sensor network. With low routing cost, good adaptability and multipath, the algorithm balanced energy consumption to prolong network lifetime. In terms of simulation and experiments, ant colony algorithm was proved to be suitable for finding optimal routing in wireless sensor network, thus achieving design goal of routing algorithm.
The improvement of the quality and efficiency of vehicle wireless network data transmission is always a key concern in the Internet of Vehicles(IoV).routing transmission solved the limitation of transmission distance ...
详细信息
The improvement of the quality and efficiency of vehicle wireless network data transmission is always a key concern in the Internet of Vehicles(IoV).routing transmission solved the limitation of transmission distance to a certain *** routing algorithm cannot adapt to complex traffic environment,resulting in low transmission *** order to improve the transmission success rate and quality of vehicle network routing transmission,make the routing algorithm more suitable for complex traffic environment,and reduce transmission power consumption to improve energy efficiency,a comprehensive optimized routing transmission algorithm is *** on the routing transmission algorithm,an optimization algorithmbased on road condition,vehicle status and network performance is proposed to improve the success rate of routing transmission in the *** distance difference and density are used as decision-making indicators to measure Road Side Unit(RSU)assisted *** the Ambient backscatter communication(AmBC)technology and energy collection are used to reduce the energy consumption of routing relay *** energy collection optimization algorithm is proposed to optimize the energy efficiency of AmBC and improve the energy efficiency of *** results show that the proposed routing optimization algorithm can effectively improve the success rate of packet transmission in vehicular ad hoc networks(VANETs),and theAmBC optimization algorithmcan effectively reduce energy consumption in the transmission *** proposed optimization algorithm achieves comprehensive optimization of routing transmission performance and energy efficiency.
暂无评论