Deep Learning for Anomaly Detection in Large-Scale Industrial Data

2018 5th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON), Journal Year: 2023, Volume and Issue: unknown, P. 1551 - 1556

Published: Dec. 1, 2023

Industrial data has increased significantly in the emerging data-driven age, and it often contains abnormalities that could point to crucial system faults or inefficiencies. The complexity high dimensionality of provide special hurdles for anomaly identification such large-scale settings. In this study, a robust deep learning framework detection is presented, one can function with large complex datasets are common industrial applications. To capture temporal spatial relationships present sensor data, makes use sophisticated neural network designs, as convolutional networks (CNNs) recurrent (RNNs). suggested model learns underlying structure using unsupervised learning, which allows recognize variations may indicate possible abnormalities. An extensive dataset used evaluate system's effectiveness, results reveal performs better than conventional machine techniques terms both computing efficiency accuracy. flexibility scalability concept reinforced by its implementation across many sectors, further demonstrates adaptability. study not only advances theoretical understanding mechanisms but also provides industry practitioners useful tool ensure safety dependability operations face increasing complexity.

Language: Английский

---

AI and ML for Enhancing Crop Yield and Resource Efficiency in Agriculture

2018 5th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON), Journal Year: 2023, Volume and Issue: unknown, P. 1605 - 1610

Published: Dec. 1, 2023

In this study, we investigate how AI and ML might revolutionize the agricultural industry, particularly with regard to increasing crop output while decreasing input costs. Applying technology has promise in a society struggling population increase, climate change, resource constraints. This study highlights practical advantages of agriculture via well-crafted research process, including data gathering, model creation, assessment. The results show that models are useful for forecasting yields, identifying illnesses, allocating resources efficiently, assisting farmers decision-making based on empirical evidence. Results like highlight importance these technologies advancing goals efficiency, sustainability, food safety. Additionally, acknowledges significance addressing ethical problems deployment, guaranteeing equal access advancements. We should expect see more into cutting-edge methods, Internet Things (IoT) integration, accessible tools subsistence as go further use sector. full designing resilient, productive, sustainable future requires collaborative efforts across stakeholders. struggle feed globe protecting its resources, shines bright light optimism.

Language: Английский

---

Low Earth Orbit (LEO) Satellite Networks: A New Era in Global Communication

2018 5th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON), Journal Year: 2023, Volume and Issue: unknown, P. 1563 - 1568

Published: Dec. 1, 2023

Global communication paradigms have undergone a dramatic transition with the introduction of Low Earth Orbit (LEO) satellite networks. This study provides an in-depth analysis architectural and technical developments in low-orbit constellations that allow high-speed, low-latency, globally accessible systems. Real-time applications were previously hindered by latency are made possible LEO networks, which use lower orbital altitudes to make significant improvements signal transmission times when compared typical geostationary satellites. The explores complex network topologies provide smooth worldwide coverage, including ground station interconnection handover techniques. Insights into these networks' capacity ubiquitous internet access underserved rural areas provided further their influence on closing digital divide. Furthermore, assesses difficulties deploying satellites suggests strategic frameworks for long-term operation. These obstacles include spectrum management debris mitigation. research emphasizes significance networks promoting global development effective disaster response while also extending socio-economic ramifications extensive deployment. makes claim potential completely transform international future synthesizing existing new trends.

Language: Английский

---

Automated Kidney Stone Detection using Deep Learning Technique

Published: Feb. 22, 2024

Convolutional Neural Networks(CNN) are created to work mostly on the image datasets and have revolutionized classification object detection by introducing versatile architectures which can be modified according requirements needed with help of modifiable hyperparameters like architecture specifications, batch size, kernel, stride loss function, learning rate etc.,. The use ResNet introduces residual pathways accelerates weight convergence compared traditional neural networks other CNN AlexNet, LeNet, GoogLeNet effectively preserves much patterns informations contained in images giving a good accuracy almost all cases considering dataset quality task complexity. In this proposed work, CT Kidneys divided into train (1453 images) test (346 images), encompassing both stones non-stone cases. Employing ResNet50 meticulously configured tailored preprocessing methods made learn data specific number epochs suggested rate. After training model for 50 epochs, applied detect stone's presence set achieved an 93%. limitations conventional machine models tasks Support Vector Machine, Logistic Regression RandomForest demonstrates their challenges capturing complex features, often results lower accuracy. And deep must need Graphics Processing Unit (GPU) reduce computation time memory management.

Language: Английский

---

Transforming Healthcare With AI and Machine Learning

Advances in healthcare information systems and administration book series, Journal Year: 2024, Volume and Issue: unknown, P. 53 - 82

Published: Nov. 27, 2024

This survey paper gives an insight about the application of artificial intelligence (AI) and machine learning (ML) in healthcare industry. With a broad focus, it discusses present state, approaches, advancements AI ML health care explores process accumulating structuring data, dataset standardization, data quality assessment, prognosis, clinical decision support systems, operations' efficiency, population dynamics, fraudulence identification, revenue cycle, patient participation, telemonitoring, individualized treatment, ethical issues, real-world examples applications developments. comprehensive highlights transformative potential healthcare, emphasizing need for continuous research, practices, robust management to harness these technologies effectively.

Language: Английский

---

Deep Learning Approaches for Feature Extraction in Big Data Analytics

2018 5th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON), Journal Year: 2023, Volume and Issue: unknown, P. 964 - 969

Published: Dec. 1, 2023

In the context of big data analytics, this study examines use algorithms based on deep learning for feature extraction. Traditional methods usually have trouble sifting through complexity and volume to find important elements. We investigate application auto encoders, transformer-based models, convolutional neural networks (CNNs), recurrent (RNNs) address problem. Our comprehensive review existing literature compares these techniques with traditional highlights their adaptability large-scale datasets. The efficacy precision methodologies are demonstrated by empirical investigations conducted authentic datasets across a range disciplines, including but not limited time-series analysis, picture identification, natural language processing. Despite challenges like computational requirements model interpretability, our findings indicate that learning-based extraction holds significant promise enhancing leading valuable insights discoveries in various fields.

Language: Английский

---

Ensemble Learning Approaches for Big Data Classification Tasks

2018 5th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON), Journal Year: 2023, Volume and Issue: unknown, P. 1545 - 1550

Published: Dec. 1, 2023

Ensemble learning has emerged as a potent method for improving prediction accuracy in Big Data classification tasks. This paper presents comprehensive study of ensemble techniques, specifically focusing on their applicability and performance handling vast complex datasets. A detailed exploration various methodologies such bagging, boosting, stacking is conducted, with particular emphasis adaptability to challenges. The further delves into novel hybrid models that synergize multiple algorithms capitalize individual strengths. quantitative analysis performed several benchmark datasets evaluate the these strategies against standalone classifiers. results indicate significant enhancement accuracy, robustness, error reduction, underlining efficacy approaches domain. also introduces framework dynamic selection, which intelligently chooses subset tailored specific characteristics dataset question. showcases potential methods evolving data landscapes, making them invaluable tools practitioners. implications findings suggest paradigm shift predictive modeling, steering future research towards more adaptive, scalable, accurate systems.

Language: Английский

---

Prediction of Concrete Strength with Artificial Neural Networks Enhanced by Nano-Silica

2018 5th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON), Journal Year: 2023, Volume and Issue: unknown, P. 959 - 963

Published: Dec. 1, 2023

The investigation is centered to create a methodology utilizing an artificial neural network (ANN) forecast concrete's crushing resistance at the 28-day mark incorporating nanosilica offering as alternative traditional cement. Nanosilica commonly used construction material known for its ability enhance strength of concrete. Nevertheless, accurately predicting this requires intricate computations, consumes time, and demands expertise. To address this, AI model was created using machine learning algorithms, trained on dataset comprising experimental compressive data nanosilica-concrete mixtures, along with additional information particle properties. model's efficacy assessed independent dataset, which demonstrated nanosilica's high precision. objective research use ANN technique establish link between various input factors resultant in nanosilica-containing utilized study sourced from existing literature.

Language: Английский

---

Neuro-Symbolic AI: Integrating Symbolic Reasoning with Deep Learning

2018 5th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON), Journal Year: 2023, Volume and Issue: unknown, P. 1587 - 1592

Published: Dec. 1, 2023

Neuro-symbolic artificial intelligence (AI) stands at the frontier of machine learning by amalgamating interpretability and structured knowledge representation symbolic reasoning with adaptive capabilities deep neural networks. This paper presents a comprehensive framework for neuro-symbolic integration, outlining harmonized architecture that leverages strengths both domains. The proposed system utilizes AI to impose structural constraints inject domain into process, enhancing models. Concurrently, it capitalizes on proficiency in handling high-dimensional, noisy data, enabling components operate beyond discrete, well-defined environments. is validated through series experiments demonstrating enhanced performance tasks requiring complex reasoning, generalization, transfer. showcases significant reduction data dependency model training, increased decision-making robustness noise ambiguity. integration marks stride towards development systems advanced cognitive abilities, akin human-like understanding reasoning. concludes discussion implications advancing field its potential transform future applications.

Language: Английский

---

Reactions of a high explosive under low intensity impact with adjustable amplitude and duration

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Nov. 30, 2024

Laboratory testing with adjustable loading amplitudes and durations remains the primary method for assessment of safety explosives under either launch or penetration environment. In this study, a novel impact laboratory equipment ranging from 0.1 to 1.0 GPa pulse 1 8 ms is established. It was used investigate 2,4-dinitroanisole (DNAN)-based melt-cast explosive subjected in scenarios. The explosive's response depends not only on characteristics (peak pressure maximum rate rise) but also confinements explosives. ignition events exhibited some randomness. A logistic regression analysis utilized analyze such events. This can predict DNAN-based high accuracy, which demonstrates effectiveness method. effect accuracy investigated.

Language: Английский

---