Today, we are constantly surrounded by vast amounts of data, a trend that is expected to grow significantly over the next decade. The abundance of data presents challenges for thorough analysis and extraction of valua...
详细信息
Sensing body vital signals, such as ECG, EEG, EMG, and EOG (ExG), is crucial for human health monitoring. To achieve accurate signal acquisition, sensing frontends are expected to have high input impedance $(Z_{1\math...
详细信息
ISBN:
(数字)9798331517458
ISBN:
(纸本)9798331517465
Sensing body vital signals, such as ECG, EEG, EMG, and EOG (ExG), is crucial for human health monitoring. To achieve accurate signal acquisition, sensing frontends are expected to have high input impedance $(Z_{1\mathrm{N}} > 10\mathrm{M}\Omega)$ to reduce signal attention, wide input linear range $(IR > 1 V_{PP})$ to better tolerate large artifacts and interferences, and low power consumption (<10µW) to prolong battery life. Recently, direct digitizing sending frontends with high resolution and low power consumption have been well developed, demonstrating their potential to meet these requirements. However, the existing state-of-the-art still faces challenges with either limited Z IN or limited IR.
This paper proposes a new hybrid routing algorithm, ReLeVR, for vehicular ad hoc networks (VANETs). Our algorithm aims to efficiently solve the problem of routing messages from vehicles to specific geographical locati...
详细信息
This paper proposes a new hybrid routing algorithm, ReLeVR, for vehicular ad hoc networks (VANETs). Our algorithm aims to efficiently solve the problem of routing messages from vehicles to specific geographical locations over VANETs. We assume that a VANET consists of vehicles and roadside units (RSUs). First, we propose a simple strategy to find the optimal locations for RSUs with the objective of minimizing their number. Our strategy computes RSU locations using available prior traffic information. Then we use Q-learning, a reinforcement learning algorithm, to learn from traffic flow patterns and compute a routing policy for the vehicles. This policy determines which grid the message should be sent to in the next step along the way for each location in the city. The routing policy is broadcast to all of the vehicles in the VANET. We demonstrate through simulations on real traffic data that our algorithm outperforms an older algorithm GPSR and a recent Q-learning-based algorithm, QGrid G in terms of metrics like the delivery ratio and delay. We observe that the number of RSUs used by our algorithm is significantly lower than that of a recent algorithm, QTAR.
This paper presents a novel state-space clustering technique specifically designed for Reinforcement Learning (RL) problems with discrete state and action spaces. Building on this clustering methodology, we propose a ...
详细信息
The global digitization of most activities of corporate and non-corporate bodies has come to stay. The interconnection of physical object "things" embedded with different technologies such as sensors, softwa...
详细信息
The global digitization of most activities of corporate and non-corporate bodies has come to stay. The interconnection of physical object "things" embedded with different technologies such as sensors, softwares and other technologies to connect and allow for data exchange with different systems and devices over the internet is known as the Internet of Things (IoT). To guarantee lower expenses and increased operational effectiveness, Data-driven insights for prompt decision-making, complete, remote asset and resource management, Prescriptive, predictive, and real-time insights along with enhanced end-user satisfaction Industrial Internet of Things (IIoT) became a given. Implementing machine learning (ML) algorithms for intrusion detection in IIoT systems entails numerous hurdles, such as data imbalance and adversarial attacks. And different ML algorithms handle data imbalance and adversarial attacks for intrusion detection differently. Therefore, there is a need for comparative studies on ML algorithms to be used as a tool to help prevent unauthorised intrusion into various networks. In this research, an experimental approach has been used to compare contemporary ML algorithms such as Linear SVC, random forest RF, XGBoost, decision tree (DT), logistic regression (LR) on IIoT datasets to detect intrusion. Data exploration, Feature engineering, selection and data partition were done at a rate of 80-20 % ratio for the train set and test set respectively. Performance evaluation was carried out using accuracy, recall, precision and F1-score and a confusion matrix was plotted. The performance evaluation shows the accuracy of 100%, 99.99%, 99.35%, 97.49% and 97.27% for RF, XGBoost, DT, Linear SVC and LR algorithms respectively the test set evaluation. RF outperformed the other algorithm that was experimented.
Spine fractures pose an important health concern that requires a quick diagnosis to protect people from any long-term complications. This paper aims to propose an automatic prediction system that uses deep transfer le...
详细信息
Breastfeeding involves a complex coordination of swallowing, breathing, and sucking, with the infant's sucking proficiency being crucial for adequate nutrient intake. However, real-time assessment of milk intake i...
详细信息
Stance Detection (SD) in the context of social media has emerged as a prominent area of interest with implications for social, business, and political applications, thereby garnering escalating research attention with...
详细信息
Stance Detection (SD) in the context of social media has emerged as a prominent area of interest with implications for social, business, and political applications, thereby garnering escalating research attention within the realm of Natural Language Processing (NLP). The inherent subtlety, nuance, and complexity of texts procured from online platforms via crowd-sourcing pose challenges for SD algorithms in accurately discerning the author’s stance. Particularly, the inclusion of sarcastic and figurative language drastically impacts the performance of SD models. This paper addresses this challenge by employing sarcasm detection intermediate-task transfer learning tailored for SD. The proposed methodology involves the fine-tuning of BERT and RoBERTa and the sequential concatenation of convolutional, bidirectional LSTM, and dense layers. Rigorous experiments are conducted on publicly available benchmark datasets to evaluate our transfer-learning framework. The performance of the approach is assessed against various State-Of-The-Art (SOTA) baselines for SD, providing empirical evidence of its effectiveness. Notably, our model outperforms the best SOTA models, achieving average F1-score gaps of 0.038 and 0.053 on the SemEval 2016 Task 6A Dataset (SemEval) and Multi-Perspective Consumer Health Query Data (MPCHI), respectively, even prior to sarcasm-detection pre-training. The integration of sarcasm knowledge into the model proves instrumental in mitigating misclassifications of sarcastic textual elements in SD. Our model accurately predicts 85% of texts that were previously misclassified by the model without sarcasm-detection pre-training, thereby amplifying the average F1-score of the model. Furthermore, our experiments revealed that the success of the transfer-learning framework is contingent upon the correlation of lexical attributes between the intermediate task (sarcasm detection) and the target task (SD). This study represents the first exploration of sarcasm detect
Concerns about security and privacy were heightened by the increasing dependence on interconnected networks, underscoring the vital role intrusion detection systems (IDS) play in protecting private data. As essential ...
详细信息
Internet of Things (IoT) devices are a fast-growing market, but they also possess a great challenge in terms of cyber-security, and therefore need a strong intrusion detection system (IDS). We present a machine learni...
详细信息
ISBN:
(数字)9798331509934
ISBN:
(纸本)9798331509941
Internet of Things (IoT) devices are a fast-growing market, but they also possess a great challenge in terms of cyber-security, and therefore need a strong intrusion detection system (IDS). We present a machine learning-based IDS that improves attack classification through effective Feature engineering and optimized Model Selection. Whereas traditional methods rely on static and well-known features, our approach offers novel ones tcp. payload and tcp. options to improve your detection of injection and vulnerability attacks. Our results demonstrate that Random Forest outperforms the convolutional models while achieving the highest accuracy of 99.07% (6-class) and 98.53% (I5-class) in five models evaluated on the Edge-IIoTset dataset. Using SMOTE to handle class im-balance and incorporating network features related to behavior, the method is able to better detect minority class general attacks. Building upon that foundation, we utilized behavioral feature extraction, data balancing, and deep learning optimization techniques that top centralized IDS models use to improve accuracy and resilience across our framework.
暂无评论