In the distributed machinelearning scenario, we have Split learning (SL) and Federated learning (FL) as the popular techniques. In SL, the model is split between the clients and the server for sequential training of ...
详细信息
ISBN:
(纸本)9783031235986;9783031235993
In the distributed machinelearning scenario, we have Split learning (SL) and Federated learning (FL) as the popular techniques. In SL, the model is split between the clients and the server for sequential training of clients, whereas in FL, clients train parallelly. The model splitting in SL provides better overall privacy than FL. SplitFed learning (SFL) combines these two popular techniques to incorporate the model splitting approach from SL to improve privacy and utilize the generic FL approach for faster training. Despite the advantages, the distributed nature of SFL makes it vulnerable to data poisoning attacks by malicious participants. This vulnerability prompted us to study the robustness of SFL under such attacks. The outcomes of this study would provide valuable insights to organizations and researchers who wish to deploy or study SFL. In this paper, we conduct three experiments. Our first experiment demonstrates that data poisoning attacks seriously threaten SFL systems. Even the presence of 10% malicious participants can cause a drastic drop in the accuracy of the global model. We further perform a second experiment to study the robustness of two variants of SFL under the category of targeted data poisoning attacks. The results of experiment two demonstrate that SFLV1 is more robust than SFLV2 the majority of times. In our third experiment, we studied untargeted data poisoning attacks on SFL. We found that untargeted attacks cause a more significant loss in the global model's accuracy than targeted attacks.
Deep learning comes under machinelearning that accomplishes more power and flexibility by learning to present different concepts or relations of real world to simpler concepts. We use Deep learning fundaments in this...
详细信息
For many developing countries like India the role of the agricultural sector is very significant. Chili has a very high economic value. India is ranked 5th in the production and 1st in exporting chili. Chili is one of...
详细信息
Modern engineering science and the area of medicine, which are traditionally seen as two seemingly opposed poles of academic brilliance, are now in close contact thanks to the discipline of biomedical engineering. Cur...
详细信息
Identifying and locating objects in images and videos, including elements like traffic signs, vehicles, buildings, and people, constitutes a fundamental and demanding task in computer vision, known as object detection...
详细信息
ISBN:
(纸本)9783031821523;9783031821530
Identifying and locating objects in images and videos, including elements like traffic signs, vehicles, buildings, and people, constitutes a fundamental and demanding task in computer vision, known as object detection. Due to the higher computing complexity of this technique and the large amount of data carried by the video signal, it is nearly impossible for ordinary general-purpose processors GPPs or CPUs to run these techniques in real-time, especially for embedded systems applications. Therefore, special hardware that can acquire, control, or execute in parallel is required. These specialized hardware systems include Digital Signal Processors DSPs, Field Programmable Gate Arrays FPGAs, Visual processing Units VPUs, Tensor processing Units TPUs, Neural processing Units NPUs or Graphics processing Units GPUs. This work presents the benefits of accelerating traditional object detection methods on a high-end embedded system, the Jetson Nano Developer Kit. This single computer board is equipped with the Tegra K1 System on Chip SoC, which is composed of a quad-core ARM A15 and 192 cores of Kepler-embedded GPU. Computing acceleration was ensured via the use of the CUDA OpenCV library for both the Histogram of Oriented Gradients HOG and the Haar Cascade Classifier. For VGA resolution, results reveal that the GPU implementation on this embedded system is 1.4x faster than the CPU for the HOG method and 2x for the Haar Cascade Classifier method.
Medical imaging helps doctors make more informed treatment decisions by providing a visual representation of the anatomy that can't be gleaned from a physical exam alone. Medical diagnosis relies heavily on the vi...
详细信息
Sepsis is a complex and heterogeneous syndrome that continues to pose significant challenges to healthcare globally, particularly in realm of Al-powered early detection. The increasing demand for Intensive Care Unit (...
详细信息
Sign language recognition is now possible because of the advancements in processors, computer vision, imageprocessing, machine and deep learning techniques. This development has made it feasible to interpret sign lan...
详细信息
The proceedings contain 36 papers. The special focus in this conference is on Web Information Systems Engineering. The topics include: Effective Transparent Monitoring of Personal Data;iCNN-LstM: An Incremental C...
ISBN:
(纸本)9789819614820
The proceedings contain 36 papers. The special focus in this conference is on Web Information Systems Engineering. The topics include: Effective Transparent Monitoring of Personal Data;iCNN-LstM: An Incremental CNN-LstM Based Ransomware Detection System;a Mini Review on Purchaser Security in the Metaverse: Challenges and Solutions;low-Resource Dataset Synthetic Generation for Hate Speech Detection;Web Open Data to SDG Indicators: Towards an LLM-Augmented Knowledge Graph Solution;leveraging Sentence-Transformers to Overcome Query-Document Vocabulary Mismatch in Information Retrieval;time Distance Aware for Multi-component Graph Collaborative Filtering;scientific Documents Recommendation Based on Graph Convolutional Network;semantic Communication of images Using image Generation and image Captioning Models;leveraging Optimization Techniques for Effective Arabic Query Expansion;DURLLCON: Deep Reinforcement learning for URLLC Optimization in Multi-edge Networks;FMM-RNS: A Fast HMM Map Matching Method Based on Road Network Simplification;Context-Aware Selection of machinelearning as a Service (MLaaS) in IoT Environments;GraphTFD: A Fraud Detection System Based on Graph Transformer;MLGE-AC-UFD: Multi-level Graph Embedding and Approximate Computation for Unsupervised Fraud Detection;SeCORE: Quantitative Security Assurance and Evaluation Platform;developing Geospatial Web Applications Using Question Answering Engines, Knowledge Graphs and Linked Data Tools;A Visual Query Builder for DBpedia;FL-PPELA: Partial Parameter Enhancement and Local Adaptive Aggregation for Personalized Federated learning;ORCPM: An Online Regional Core pattern Mining System;MTRM: A Web-Miner Multi-Threshold Mining Co-location patterns to Mitigate Redundancy;constrained Path Optimization on Time-Dependent Road Networks;prompt strategies for Sarcastic Meme Detection: A Comparative Analysis;Deepfake Detection in Cancer Medical Imaging Using CNN Architectures;strengthening Cybersecurity: The Influence o
Lung Cancer has become serious health problem which affects all age groups from children to old age persons. An early-stage prognosis and diagnosis of lung cancer can save many humans from major causalities. X-rays, U...
详细信息
ISBN:
(纸本)9798350377002
Lung Cancer has become serious health problem which affects all age groups from children to old age persons. An early-stage prognosis and diagnosis of lung cancer can save many humans from major causalities. X-rays, Ultrasounds and CT (Computerized Tomography) are further diagnostic tools. The goal of computer science, often known as machinelearning (ML), is to develop algorithms that let computers pick up new practices from examples seen in the real world. There exist several machinelearning methods which are used to categorize lung tumors which finally develop into lung cancer. Precision, Recall and F1 Score parameters are used for testing the model's capability. Using a test dataset of 200 epochs, the CNN system obtains a high accuracy of approx. 98.87% in identifying normal and cancerous cells. This study examines the use of optimized feature selection and imageprocessing in the identification of initial phase of lung cancer. The rigorous evaluation is done on large dataset while taking in consideration healthy and malignant cells. The study includes training, pre-processing, model building and performance analysis of deep learning algorithm which gives better results in case of forecasting and curing lung cancer at the initial stage so that risk of causalities can be minimized as compared with conventional techniques. This paper also examines the benefits of machinelearning techniques in healthcare, cancer prognosis, and detection. Many researchers have concentrated on creating cancer prediction systems that incorporate classification algorithms for accurate results and use supervised learning approaches in machinelearning. The importance of deep learning and related algorithms in the healthcare industry is emphasized. The research recommends enhancing and expanding the lung cancer system through deep learning techniques in order to improve the accuracy of both lung cancer identification and prediction. With a view to implementing deep learning techniques
暂无评论