With the integration of large-scale distributed PV into the distribution network, the PV penetration rate is increasing, and the imbalance between the PV output and the load leads to the reverse current in the line, c...
详细信息
The proceedings contain 71 papers. The topics discussed include: extracting software design from text: a machine learning approach;deep reinforcement learning based actor-critic framework for decision-making actions i...
ISBN:
(纸本)9781665440769
The proceedings contain 71 papers. The topics discussed include: extracting software design from text: a machine learning approach;deep reinforcement learning based actor-critic framework for decision-making actions in production scheduling;model-based test case generation approach for mobile applications load testing using OCL enhanced activity diagrams;deep learning achievements and opportunities in domain of electronic warfare applications;a proposed approach to secure automated teller machine-based financial transactions;astronomical image denoising based on convolutional neural network;an improved deep learning model for early fire and smoke detection on edge vision unit;long-term person re-identification model with a strong feature extractor;satellite imagery road segmentation using a dual network approach with enhancement blocks;multi-limb split learning for tumor classification on vertically distributed data;deep learning methodologies for human activity recognition;unlocking the public perception of COVID-19 vaccination process on social media;and identification of a new topology to enhance the impedance extraction in microfluidic systems.
The likelihood of unanticipated node failures in large-scale parallel computers increases with growing numbers of nodes. Furthermore, global reduction operations become major bottlenecks due to their limited parallel ...
详细信息
ISBN:
(纸本)9798350364613;9798350364606
The likelihood of unanticipated node failures in large-scale parallel computers increases with growing numbers of nodes. Furthermore, global reduction operations become major bottlenecks due to their limited parallel scalability. The Preconditioned Conjugate Gradient (PCG) method faces these challenges.
Euclidean distance (ED) calculation is essential in wireless communication systems and signal processing applications. This paper presents a low-complexity, high-throughput method for computing Euclidean distances. Th...
详细信息
ISBN:
(数字)9798350359091
ISBN:
(纸本)9798350359107;9798350359091
Euclidean distance (ED) calculation is essential in wireless communication systems and signal processing applications. This paper presents a low-complexity, high-throughput method for computing Euclidean distances. The central idea is to utilize the in-memory computing (IMC) technique, performing computations via an array of memristor devices. Specifically, multiplications and additions are carried out using the inherent properties of memristor devices, following Ohm's law and Kirchhoff's current law. This innovative approach boosts flexibility through memristor programming and enhances processing efficiency by reducing computational complexity and latency, surpassing conventional implementation methods.
Edge computing enhances task reliability by employing redundant task executions across edge nodes. Conventional decentralized task offloading strategies, based on heuristics and game theory, either focus on optimizati...
详细信息
Edge computing has enabled a large set of emerging edge applications by exploiting data proximity and offloading computation-intensive workloads to nearby edge servers. However, supporting edge application users at sc...
详细信息
ISBN:
(数字)9781665471770
ISBN:
(纸本)9781665471770
Edge computing has enabled a large set of emerging edge applications by exploiting data proximity and offloading computation-intensive workloads to nearby edge servers. However, supporting edge application users at scale poses challenges due to limited point-of-presence edge sites and constrained elasticity. In this paper, we introduce a densely-distributed edge resource model that leverages capacity-constrained volunteer edge nodes to support elastic computation offloading. Our model also enables the use of geo-distributed edge nodes to further support elasticity. Collectively, these features raise the issue of edge selection. We present a distributed edge selection approach that relies on client-centric views of available edge nodes to optimize average end-to-end latency, with considerations of system heterogeneity, resource contention and node churn. Elasticity is achieved by fine-grained performance probing, dynamic load balancing, and proactive multi-edge node connections per client. Evaluations are conducted in both real-world volunteer environments and emulated platforms to show how a common edge application, namely AR-based cognitive assistance, can benefit from our approach and deliver low-latency responses to distributed users at scale.
Currently, large-scale model have transitioned from a period of algorithm research to a period of industry application. This paper presents a comprehensive review of large-scale model deployment modes. It provides an ...
详细信息
ISBN:
(纸本)9798350364279;9798350364262
Currently, large-scale model have transitioned from a period of algorithm research to a period of industry application. This paper presents a comprehensive review of large-scale model deployment modes. It provides an overview of the various deployment mode, their advantages and disadvantages. We delve into the critical factors that must be taken into account when selecting a deployment mode for large-scale model. We present the various deployment modes available for large-scale models, including centralized deployment, distributed deployment, and edge computing deployment. We particularly focus on edge computing deployment mode and analyze evolving trends in large-scale model deployment. We specially conducted an analysis on converged infrastructure and edge AI of large-scale model. Furthermore, we provide insights into the main challenges and future research directions in this field.
Solving the system of linear algebraic equations (LAE) is the most well known and probably the most important of all numerical computations involving real numbers. Researchers have been committed to developing distrib...
详细信息
ISBN:
(数字)9781665471770
ISBN:
(纸本)9781665471770
Solving the system of linear algebraic equations (LAE) is the most well known and probably the most important of all numerical computations involving real numbers. Researchers have been committed to developing distributed algorithms to solve such systems for a long time. However, traditional distributed algorithms have serious security risks when the coefficients and the solution are of great value. To address the privacy issue, we propose a new secure distributed outsourcing protocol for solving large-scale LAE systems. Specifically, we give an algorithm for generating a generic repeatedly jointly strongly connected sequence, for the first time as far as we know. Then we embed the matrix masking technique in our distributed system requiring multiple rounds of iteration while keeping the correctness of convergence. In addition, for the first time, we give a method for discriminating by the agents under masking whether the plaintext approximate solution corresponding to the current masked solution satisfies a predetermined constraint. Finally, we give the experimental result to show the practicality of our new protocol.
Offloading software modules to the edge/cloud can enhance a robot's capabilities by leveraging massive computing power. However, determining which software module should be offloaded and scheduled to which robot/e...
详细信息
ISBN:
(纸本)9798350377712;9798350377705
Offloading software modules to the edge/cloud can enhance a robot's capabilities by leveraging massive computing power. However, determining which software module should be offloaded and scheduled to which robot/edge/cloud node is a challenging task, particularly for robot fleets with diverse tasks. In this paper, we tackle the software scheduling problem and introduce a taxonomy to categorize software modules and classify their applicability and requirements for offloading. Additionally, by using prior measurements, we model the compute cluster and formalize software scheduling as a multi-objective optimization problem which we tackle with a genetic algorithm. To evaluate our approach with a challenging setup, we build a mobile manipulation task using open-source frameworks and libraries in the Robot Operating System (ROS2) community in simulation as well as a mildly simplified real-world variant. Our evaluation shows significant improvements compared to the built-in scheduler of Kubernetes (K8s) regarding robotic specific metrics such as the rate of missed cycle time in both simulated and real-world experiments.
We consider the problem of cost-effectively mapping a swarm of soft real-time stream processing applications with moldable-parallel tasks to multicore resources in the device-edge-cloud continuum, consisting of mobile...
详细信息
ISBN:
(纸本)9798350366495;9798350366488
We consider the problem of cost-effectively mapping a swarm of soft real-time stream processing applications with moldable-parallel tasks to multicore resources in the device-edge-cloud continuum, consisting of mobile devices, edge resources and cloud resources. We leverage flexibility from different parallelization degrees and frequency levels (DVFS) for the tasks, keeping application throughput constraints and communication bandwidth limitations while minimizing overall cost (including device/edge resource energy and cloud resource renting). We present two offline algorithmic solutions with a global view of the environment: an integer linear program (ILP) extending the crown scheduling approach for multi-layer distributedsystems and a greedy heuristic algorithm. Our experimental evaluation for several real-world and synthetic scenarios shows that the time required for solving the scheduling problem to cost-optimality by the ILP is feasible for nontrivial scenarios. The heuristic achieves about 12% worse cost efficiency on average, yet operates much faster (by 1-2 orders of magnitude), allowing to scale up the problem size more than the ILP approach.
暂无评论