Generative Deep Neural Networks for Estimating Hypervariability in Hepatitis B and C Virus Genomes DOI Creative Commons

Sharmeen Saqib,

Zilwa Mumtaz, Hania Ahmed

et al.

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 21, 2024

Abstract Hepatitis B virus (HBV) and C (HCV) have always remained a greater global concern. Approximately 1.3 million deaths occur each year due to HBV HCV. Due the diverse genotypes drug resistance, diagnostic challenges are being faced treat these viruses. Therefore, success ratio of antiviral therapies has been decreasing with time in last few decades. By deep learning predictive model, pattern evolution hypervariable regions HCV genes can be foreseen. In HCV, region is Envelope glycoprotein (E2) gene, while HBV, it includes S1 S2 genes. Generative models used for evolutionary studies, but application limited viral research predicting evolving The Long Short-Term Memory (LSTM) model represented satisfactory outcome sequences that might great help diagnosis vaccine design. We collected data from databases like NCBI BVBRC. Our proposed LSTM generative was trained on 1500 present 7 10 HBV. Apart traditional Recurrent Neural Network (RNN), our not only generates sequence also learns develops relationship between various parts virus’s genetic code. this study, three were compared, Simple RNN, 1-Dimensional Convolutional (ConV1d) (LSTM). Among three, demonstrated least error rate highest efficiency accuracy. While simple RNN ConV1d illustrated relatively higher lower gained reading long dependencies, hence, efficient at handling sequential along preventing conventional issue losing important information data, which happens frequently ConV1d.

Language: Английский

Multimodal data fusion for geo-hazard prediction in underground mining operation DOI Creative Commons
Ruiyu Liang, Chengguo Zhang, Chaoran Huang

et al.

Computers & Industrial Engineering, Journal Year: 2024, Volume and Issue: 193, P. 110268 - 110268

Published: June 5, 2024

Geohazard prediction is one of the most important and challenging tasks in underground mining. It still remains difficult to improve accuracy make it compatible with ever-increasing data mining, especially when are sparsely allocated a large-scale mining environment. This study introduces an innovative multimodal fusion approach for geohazard address this challenge. By incorporating visual model as novel modality using interpolated rock mass rating cross-complementary factor, framework enhances effectiveness fusion. Specific machine learning models were used validated (e.g., neural networks, SVM, KNN, etc.) proposed fusion, addressing challenges posed by scattered multidimensional data, which generally have weak spatial connections across diverse datasets. In detail, enhance connection among datasets, paper leverage digitalised gridded CAD file-based foundational carrier, new modality, facilitate establishment robust internal routine data. Additionally, aligned connections, improving information-orientated Then, validate efficiency framework, we process integrate two different from case mine. Performance tested nine combinations, originating Finally, through comprehensive cross-validation, significantly improves stability at mine site scale, high low False-Negative rate.

Language: Английский

Citations

4
In-situ evaluation of hole quality and cutting tool condition in robotic drilling of composite materials using machine learning DOI Creative Commons
Stephen Lee, Patrick G. Mongan, Ahmad Farhadi

et al.

Journal of Intelligent Manufacturing, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 19, 2025

Abstract The massive adoption of industrial robots in the manufacturing sector has significantly increased automation installation and inspection procedures, particularly benefiting aerospace industry, where large volumes holes are drilled each aircraft. However, mechanical drilling remains challenging when dealing with composite materials due to their inherently heterogeneous structure. This work presents a novel approach for in-situ hole quality utilising integrated sensor data an robotic drill, combined machine learning model. Additionally, classification evaluating is proposed. study employed KUKA robot, fitted multifunctional end-effector, drill material used applications. An ensemble neural network (ENN) model, which combines artificial genetic algorithm, was assess these holes. model specifically developed tested on machined relate process input parameters torques quality. predictions were validated six unseen datasets, five predicted accurately. A full factorial conducted using analysis variance (ANOVA) investigate relationship between tool condition torque. results ANOVA show that largest contributor method proposed this work, allows real-time monitoring quality, potential improve productivity components while ensuring high-quality end products.

Language: Английский

Citations

0
Analytical Model of CFRP Cutting Mechanics with Strain Rate Effect DOI Creative Commons
Zhenghui Lu, Xiaoliang Jin

International Journal of Mechanical Sciences, Journal Year: 2025, Volume and Issue: unknown, P. 110206 - 110206

Published: March 1, 2025

Language: Английский

Citations

0
Method of bed exit intention based on the internal pressure features in array air spring mattress DOI Creative Commons
Fanchao Meng, Teng Liu, Chuizhou Meng

et al.

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

Published: Nov. 8, 2024

With the population ages, many patients are unable to receive comprehensive care, leading an increase in hazardous incidents, particularly falls occurring after getting out of bed. To address this issue, paper proposes a method for recognizing bed-exit intentions using array air spring mattress. The integrates convolutional neural networks with feature point matching techniques identify both global and local features spring. For features, one-dimensional focal loss network (1D-FLCNN) model is employed classify eight internal pressure time series determine status based on features. distribution matrix extracted represent spatial postures. Euclidean distance utilized measure similarity between these matrices match Finally, recognition results from types combined logical OR operation produce final result. Experimental validation confirms that proposed greatly improves anti-interference capability effectively avoids problem non-recognition due body position external environment.

Language: Английский

Citations

2
Tool Wear Classification in Chipboard Milling Processes Using 1-D CNN and LSTM Based on Sequential Features DOI Creative Commons
Jarosław Kurek, Elżbieta Świderska, Karol Szymanowski

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(11), P. 4730 - 4730

Published: May 30, 2024

The paper presents the comparative analysis of Long short-term memory (LSTM) and one-dimensional convolutional neural networks (1-D CNNs) for tool wear classification in chipboard milling processes. complexity sequence data various fields makes selecting right model very important. This research aims to show distinct capabilities performance nuances LSTM 1-D CNN models, leveraging their inherent strengths understanding temporal dependencies feature extraction, respectively. Through a series experiments, study unveils that while both models demonstrate competencies handling data, model, with its superior extraction capabilities, achieved best performance, boasting an accuracy 94.5% on test dataset. insights gained from this comparison not only help understand better, but also open door future improvements using complex challenges.

Language: Английский

Citations

1
Research on portable force-controlled machining device of CFRP and the method of thrust force controlling DOI
Fei Su,

Ziheng Zeng,

Ke Chen

et al.

Journal of Manufacturing Processes, Journal Year: 2024, Volume and Issue: 131, P. 781 - 796

Published: Sept. 24, 2024

Language: Английский

Citations

1
Online prediction of composite material drilling quality based on multi-sensor fusion DOI
Wei Liu, Jiacheng Cui, Yongkang Lu

et al.

Journal of Intelligent Manufacturing, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 14, 2024

Language: Английский

Citations

1
Accurate synthesis of sensor-to-machined-surface image generation in carbon fiber-reinforced plastic drilling DOI Creative Commons
Jae Gyeong Choi,

Dong Chan Kim,

M. Chung

et al.

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 255, P. 124656 - 124656

Published: July 6, 2024

Delamination is a prevalent issue in carbon fiber-reinforced plastic (CFRP) drilling, significantly compromising the mechanical properties of material. Considering that delamination can impact long-term durability final products, it essential for operators to promptly identify it. This paper proposes machined surface image generation model, called Sensor2Image, employs time-series force sensor data as input and generates drilled-hole images output. Sensor2Image first encodes into using Gramian angular field (GAF) method. Subsequently, applies an image-to-image translation technique generate images. The proposed model was trained evaluated experimental gathered from drilling CFRP specimens under industrial robot machining system. results demonstrated versatility practical applications, regardless factor. method offers significant advantages over existing methods through its intuitive visual representation approach. It facilitates inspection while enabling quality analysis drilled hole identification defects or irregularities may approach enhance efficiency reliability processes, particularly those involving complex factors. valuable tool optimizing process enhancing user-friendly manner.

Language: Английский

Citations

1
Layer Selection for Subtraction and Concatenation: A Method for Visual Velocity Estimation of a Mobile Robot DOI Open Access
Mustafa Can Bingöl

Bitlis Eren Üniversitesi Fen Bilimleri Dergisi, Journal Year: 2024, Volume and Issue: 13(2), P. 384 - 392

Published: June 29, 2024

Kinematic information such as position, velocity, and acceleration is critical to determine the three-dimensional state of robot in space. In this study, it aimed estimate visual linear angular velocity a mobile robot. Additionally, another aim study suitability concatenation or subtraction layer Convolutional Neural Network (CNN) that will make estimate. For these purposes, first, simulation environment was created. 9000 pairs images necessary were collected from for training. Similarly, 1000 gathered validation. Four different CNN models designed trained tested using datasets. As result test, lowest average error estimation calculated 0.93e-3m/s measured 4.37e-3rad/s. It observed results sufficient prediction according statistical analysis errors. addition, can be used instead architectures hardware-limited systems. result, robots has been achieved with framework drawn problem.

Language: Английский

Citations

0
Hyper CLS-Data-Based Robotic Interface and Its Application to Intelligent Peg-in-Hole Task Robot Incorporating a CNN Model for Defect Detection DOI Creative Commons
Fusaomi Nagata,

Ryoma Abe,

Shinichiro Sakata

et al.

Machines, Journal Year: 2024, Volume and Issue: 12(11), P. 757 - 757

Published: Oct. 26, 2024

Various types of numerical control (NC) machine tools can be standardly operated and controlled based on NC data that easily generated using widespread CAD/CAM systems. On the other hand, operation environments industrial robots still depend conventional teaching playback systems provided by makers, so it seems they have not been standardized unified like yet. Additionally, robotic functional extensions, e.g., easy implementation a learning model, such as convolutional neural network (CNN), visual feedback controller, cooperative for multiple robots, on, has sufficiently realized In this paper, hyper cutter location source (HCLS)-data-based interface is proposed to cope with issues. Due HCLS-data-based robot interface, sequence visually unifiedly described codes. addition, VGG19-based CNN model defect detection, whose classification accuracy over 99% average time forward calculation 70 ms, systematically incorporated into application handles robots. The effectiveness validity system are demonstrated through pick place task three small-sized MG400s peg-in-hole while checking undesirable defects in workpieces without any programmable logic controller (PLC). specifications PC used experiments CPU: Intel(R) Core(TM) i9-10850K CPU 3.60 GHz, GPU: NVIDIA GeForce RTX 3090, Main memory: 64 GB.

Language: Английский

Citations

0