Search for: All records

Blockchain technology, recognized for its decentralized and privacy-preserving capabilities, holds potential for enhancing privacy in contact tracing applications. Existing blockchain-based contact tracing frameworks often overlook one or more critical design details, such as the blockchain data structure, a decentralized and lightweight consensus mechanism with integrated tracing data verification, and an incentive mechanism to encourage voluntary participation in bearing blockchain costs. Moreover, the absence of framework simulations raises questions about the efficacy of these existing models. To solve above issues, this article introduces a fully third-party independent blockchain-driven contact tracing (BDCT) framework, detailed in its design. The BDCT framework features an RivestShamir-Adleman (RSA) encryption-based transaction verification method (RSA-TVM), achieving over 96% accuracy in contact case recording, even with a 60% probability of individuals failing to verify contact information. Furthermore, we propose a lightweight reputation corrected delegated proof of stake (RCDPoS) consensus mechanism, coupled with an incentive model, to ensure timely reporting of contact cases while maintaining blockchain decentralization. Additionally, a novel simulation environment for contact tracing is developed, accounting for three distinct contact scenarios with varied population density. Our results and discussions validate the effectiveness, robustness of the RSA-TVM and RC-DPoS, and the low storage demand of the BDCT framework. 

Social media9s explosive growth has resulted in a massive influx of electronic documents influencing various facets of daily life. However, the enormous and complex nature of this content makes extracting valuable insights challenging. Long document summarization emerges as a pivotal technique in this context, serving to distill extensive texts into concise and comprehensible summaries. This paper presents a novel three-stage pipeline for effective long document summarization. The proposed approach combines unsupervised and supervised learning techniques, efficiently handling large document sets while requiring minimal computational resources. Our methodology introduces a unique process for forming semantic chunks through spectral dynamic segmentation, effectively reducing redundancy and repetitiveness in the summarization process. Contrary to previous methods, our approach aligns each semantic chunk with the entire summary paragraph, allowing the abstractive summarization model to process documents without truncation and enabling the summarization model to deduce missing information from other chunks. To enhance the summary generation, we utilize a sophisticated rewrite model based on Bidirectional and Auto- Regressive Transformers (BART), rearranging and reformulating summary constructs to improve their fluidity and coherence. Empirical studies conducted on the long documents from the Webis-TLDR-17 dataset demonstrate that our approach significantly enhances the efficiency of abstractive summarization transformers. The contributions of this paper thus offer significant advancements in the field of long document summarization, providing a novel and effective methodology for summarizing extensive texts in the context of social media. 

In monitoring station observation, for the best accuracy of rumor source detection, it is important to deploy monitors appropriately into the network. There are, however, a very limited number of studies on the monitoring station selection. This article will study the problem of detecting a single rumormonger based on an observation of selected infection monitoring stations in a complete snapshot taken at some time in an online social network (OSN) following the independent cascade (IC) model. To deploy monitoring stations into the observed network, we propose an influence-distancebased k-station selection method where the influence distance is a conceptual measurement that estimates the probability that a rumor-infected node can influence its uninfected neighbors. Accordingly, a greedy algorithm is developed to find the best k monitoring stations among all rumor-infected nodes with a 2-approximation. Based on the infection path, which is most likely toward the k infection monitoring stations, we derive that an estimator for the “most like” rumor source under the IC model is the Jordan infection center in a graph. Our theoretical analysis is presented in the article. The effectiveness of our method is verified through experiments over both synthetic and real-world datasets. As shown in the results, our k-station selection method outperforms off-the-shelf methods in most cases in the network under the IC model. 

Description / Abstract: In order to effectively provide INaaS (Inference-as-a-Service) for AI applications in resource-limited cloud environments, two major challenges must be overcome: achieving low latency and providing multi-tenancy. This paper presents EIF (Efficient INaaS Framework), which uses a heterogeneous CPU-FPGA architecture to provide three methods to address these challenges (1) spatial multiplexing via software-hardware co-design virtualization techniques, (2) temporal multiplexing that exploits the sparsity of neural-net models, and (3) streaming-mode inference which overlaps data transfer and computation. The prototype EIF is implemented on an Intel PAC (shared-memory CPU-FPGA) platform. For evaluation, 12 types of DNN models were used as benchmarks, with different size and sparsity. Based on these experiments, we show that in EIF, the temporal multiplexing technique can improve the user density of an AI Accelerator Unit from 2$$\times$$ to 6$$\times$$, with marginal performance degradation. In the prototype system, the spatial multiplexing technique supports eight AI Accelerators Unit on one FPGA. By using a streaming mode based on a Mediated Pass-Through architecture, EIF can overcome the FPGA on-chip memory limitation to improve multi-tenancy and optimize the latency of INaaS. To further enhance INaaS, EIF utilizes the MapReduce function to provide a more flexible QoS. Together with the temporal/spatial multiplexing techniques, EIF can support 48 users simultaneously on a single FPGA board in our prototype system. In all tested benchmarks, cold-start latency accounts for only approximately 5\% of the total response time.