Faster Training Time Research Articles

Wi-Fi sensing gesture control algorithm based on semi-supervised generative adversarial network

A Wi-Fi-sensing gesture control system for smart homes has been developed based on a theoretical investigation of the Fresnel region sensing model, addressing the need for non-contact gesture control in household environments. The system collects channel state information (CSI) related to gestures from Wi-Fi signals transmitted and received by network cards within a specific area. The collected data undergoes preprocessing to eliminate environmental interference, allowing for the extraction of complete gesture sets. Dynamic feature extraction is then performed, followed by the identification of unknown gestures using pattern recognition techniques. An improved dynamic double threshold gesture interception algorithm is introduced, achieving a gesture interception accuracy of 98.20%. Furthermore, dynamic feature extraction is enhanced using the Gramian Angular Summation Field (GASF) transform, which converts CSI data into GASF graphs for more effective gesture recognition. An enhanced generative adversarial network (GAN) algorithm with an embedded classifier is employed to classify unknown gestures, enabling the simultaneous recognition of multiple gestures. A semi-supervised learning algorithm designed to perform well even with limited labeled data demonstrates high performance in cross-scene gesture recognition. Compared to traditional fully-supervised algorithms like linear discriminant analysis (LDA), Light Gradient Boosting Machine (LightGBM), and support vector machine (SVM), the semi-supervised GAN algorithm achieves an average accuracy of 95.67%, significantly outperforming LDA (58.20%), LightGBM (78.20%), and SVM (75.67%). In conclusion, this novel algorithm maintains an accuracy of over 94% across various scenarios, offering both faster training times and superior accuracy, even with minimal labeled data.

Application of deep learning for rice leaf disease detection in the Mekong Delta

The Mekong River Delta, the largest rice-producing region in Vietnam with an annual output of over 25 million tons, plays a vital role in ensuring food security both within the country and globally. In recent years, it has undergone significant transformation in rice cultivation, which aims to support farmers here to plant rice more effectively. However, severe weather conditions and soil degradation have negatively impacted rice growth. Additionally, rice is highly susceptible to various diseases that must be identified and prevented promptly. As a result, leveraging technology such as AI and deep learning to diagnose rice diseases based on leaf symptoms is essential. This paper utilizes an image dataset of three common rice leaf diseases—leaf smut, brown spot, and bacterial leaf blight—and applies deep learning networks (MobileNet and ResNet) to evaluate and select the best model. A diagnostic program is then developed to detect these diseases. Experimental results show that the MobileNetV3-Small model (a variant of the MobileNet network) is the most optimal, offering fast training time, high accuracy, and acceptable levels of loss and error.

Extraordinarily Time- and Memory-Efficient Large-Scale Canonical Correlation Analysis in Fourier Domain: From Shallow to Deep.

Canonical correlation analysis (CCA) is a correlation analysis technique that is widely used in statistics and the machine-learning community. However, the high complexity involved in the training process lays a heavy burden on the processing units and memory system, making CCA nearly impractical in large-scale data. To overcome this issue, a novel CCA method that tries to carry out analysis on the dataset in the Fourier domain is developed in this article. Appling Fourier transform on the data, we can convert the traditional eigenvector computation of CCA into finding some predefined discriminative Fourier bases that can be learned with only element-wise dot product and sum operations, without complex time-consuming calculations. As the eigenvalues come from the sum of individual sample products, they can be estimated in parallel. Besides, thanks to the data characteristic of pattern repeatability, the eigenvalues can be well estimated with partial samples. Accordingly, a progressive estimate scheme is proposed, in which the eigenvalues are estimated through feeding data batch by batch until the eigenvalues sequence is stable in order. As a result, the proposed method shows its characteristics of extraordinarily fast and memory efficiencies. Furthermore, we extend this idea to the nonlinear kernel and deep models and obtained satisfactory accuracy and extremely fast training time consumption as expected. An extensive discussion on the fast Fourier transform (FFT)-CCA is made in terms of time and memory efficiencies. Experimental results on several large-scale correlation datasets, such as MNIST8M, X-RAY MICROBEAM SPEECH, and Twitter Users Data, demonstrate the superiority of the proposed algorithm over state-of-the-art (SOTA) large-scale CCA methods, as our proposed method achieves almost same accuracy with the training time of our proposed method being 1000 times faster. This makes our proposed models best practice models for dealing with large-scale correlation datasets. The source code is available at https://github.com/Mrxuzhao/FFTCCA.

Prediction of Glass Transition Temperature of Polymers Using Simple Machine Learning.

Polymer materials have garnered significant attention due to their exceptional mechanical properties and diverse industrial applications. Understanding the glass transition temperature (Tg) of polymers is critical to prevent operational failures at specific temperatures. Traditional methods for measuring Tg, such as differential scanning calorimetry (DSC) and dynamic mechanical analysis, while accurate, are often time-consuming, costly, and susceptible to inaccuracies due to random and uncertain factors. To address these limitations, the aim of the present study is to investigate the potential of Simplified Molecular Input Line Entry System (SMILES) as descriptors in simple machine learning models to predict Tg efficiently and reliably. Five models were utilized: k-nearest neighbors (KNNs), support vector regression (SVR), extreme gradient boosting (XGBoost), artificial neural network (ANN), and recurrent neural network (RNN). SMILES descriptors were converted into numerical data using either One Hot Encoding (OHE) or Natural Language Processing (NLP). The study found that SMILES inputs with fewer than 200 characters were inadequate for accurately describing compound structures, while inputs exceeding 200 characters diminished model performance due to the curse of dimensionality. The ANN model achieved the highest R2 value of 0.79; however, the XGB model, with an R2 value of 0.774, exhibited the highest stability and shorter training times compared to other models, making it the preferred choice for Tg prediction. The efficiency of the OHE method over NLP was demonstrated by faster training times across the KNN, SVR, XGB, and ANN models. Validation of new polymer data showed the XGB model's robustness, with an average prediction deviation of 9.76 from actual Tg values. These findings underscore the importance of optimizing SMILES conversion methods and model parameters to enhance prediction reliability. Future research should focus on improving model accuracy and generalizability by incorporating additional features and advanced techniques. This study contributes to the development of efficient and reliable predictive models for polymer properties, facilitating the design and application of new polymer materials.

The crowd in MOOCs: a study of learning patterns at scale

ABSTRACT The increasing availability of learning activity data in Massive Open Online Courses (MOOCs) enables us to conduct a large-scale analysis of learners' learning behavior. In this paper, we analyze a dataset of 351 million learning activities from 0.8 million unique learners enrolled in over 1.6 thousand courses within two years. Specifically, we mine and identify the learning patterns of the crowd from both temporal and course enrollment perspectives leveraging mutual information theory and sequential pattern mining methods. From the temporal perspective, we find that the time intervals between consecutive learning activities of learners exhibit a mix of power-law and periodic cosine function distribution. By qualifying the relationship between course pairs, we observe that the most frequently co-enrolled courses usually fall in the same category or the same university. We demonstrate these findings can facilitate manifold applications including recommendation tasks on courses. A simple recommendation model utilizing the course enrollment patterns is competitive to the baselines with 200× faster training time.

MEDi-SOL: Multi Ensemble Distribution Model for Estimating Sleep Onset Latency.

Sleep onset latency (SOL) is an important factor relating to the sleep quality of a subject. Therefore, accurate prediction of SOL is useful to identify individuals at risk of sleep disorders and to improve sleep quality. In this study, we estimate SOL distribution and falling asleep function using an electroencephalogram (EEG), which can measure the electric field of brain activity. We proposed a Multi Ensemble Distribution model for estimating Sleep Onset Latency (MEDi-SOL), consisting of a temporal encoder and a time distribution decoder. We evaluated the performance of the proposed model using a public dataset from the Sleep Heart Health Study. We considered four distributions, Normal, log-Normal, Weibull, and log-Logistic, and compared them with a survival model and a regression model. The temporal encoder with the ensemble log-Logistic and log-Normal distribution showed the best and second-best scores in the concordance index (C-index) and mean absolute error (MAE). Our MEDi-SOL, multi ensemble distribution with combining log-Logistic and log-Normal distribution, shows the best score in C-index and MAE, with a fast training time. Furthermore, our model can visualize the process of falling asleep for individual subjects. As a result, a distribution-based ensemble approach with appropriate distribution is more useful than point estimation.

Predictions of transient vector solution fields with sequential deep operator network

The deep operator network (DeepONet) structure has shown great potential in approximating complex solution operators with low generalization errors. Recently, a sequential DeepONet (S-DeepONet) was proposed to use sequential learning models in the branch of DeepONet to predict final solutions given time-dependent inputs. In the current work, the S-DeepONet architecture is extended by modifying the information combination mechanism between the branch and trunk networks to simultaneously predict vector solutions with multiple components at multiple time steps of the evolution history, which is the first in the literature using DeepONets. Two example problems, one on transient fluid flow and the other on path-dependent plastic loading, were shown to demonstrate the capabilities of the model to handle different physics problems. The use of a trained S-DeepONet model in inverse parameter identification via the genetic algorithm is shown to demonstrate the application of the model. In almost all cases, the trained model achieved an R2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$R^2$$\\end{document} value of above 0.99 and a relative L2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$L_2$$\\end{document} error of less than 10% with only 3200 training data points, indicating superior accuracy. The vector S-DeepONet model, having only 0.4% more parameters than a scalar model, can predict two output components simultaneously at an accuracy similar to the two independently trained scalar models with a 20.8% faster training time. The S-DeepONet inference is at least three orders of magnitude faster than direct numerical simulations, and inverse parameter identifications using the trained model are highly efficient and accurate.

Tomato Leaf Disease Detection using Flask Frame Work

In recent years, plant leaf diseases has become a widespread problem for which an accurate research and rapid application of deep learning in plant disease classification is required, tomato is also one of the most important plants and seeds which are used worldwide for cooking in either dried or fresh form, tomato are a great source of protein that offer many health benefits, but there are a lot of diseases associated with tomato leaf which hinder its production. Thus, an accurate classification of tomato leaf diseases is needed to solve the problem in the early stage. A deep learning approach is proposed to identify and classify leaf disease by using public dataset of leaf image and CNN model with the open source library TensorFlow. In this project, we proposed a method to classify tomato leaf disease and to find and describe the efficient network architecture (hyper parameters and optimization methods). Moreover, after applying each architecture separately, we compared their obtained results to find out the best architecture configuration for classifying tomato leaf diseases and their results. Furthermore, to satisfy the classification requirements, the model was trained using CNN architecture check if we could get faster training times, higher accuracy and easier retraining. Deep learning is a branch of artificial intelligence. In recent years, with the advantages of automatic learning and feature extraction, it has been widely concerned by academic and industrial circles. It has been widely used in image and video processing, voice processing, and natural language processing. At the same time, it has also become a research hotspot in the field of agricultural plant protection, such as plant disease recognition. The application of deep learning in plant disease recognition can avoid the disadvantages caused by artificial selection of disease spot features, make plant disease feature extraction more objective, and improve the research efficiency and technology transformation speed. This review provides the research progress of deep learning technology in the field of crop leaf disease identification in recent years. In this project, we present the current trends and challenges for the detection of plant leaf disease using deep learning and advanced imaging techniques. We hope that this project will be a valuable resource for researchers who study the detection of plant diseases. At the same time, we also discussed some of the current challenges and problems that need to be resolved.

A comprehensive machine learning-based investigation for the index-value prediction of 2G HTS coated conductor tapes

Index-value, or so-called n-value prediction is of paramount importance for understanding the superconductors’ behaviour specially when modeling of superconductors is needed. This parameter is dependent on several physical quantities including temperature, the magnetic field’s density and orientation, and affects the behaviour of high-temperature superconducting devices made out of coated conductors in terms of losses and quench propagation. In this paper, a comprehensive analysis of many machine learning (ML) methods for estimating the n-value has been carried out. The results demonstrated that cascade forward neural network (CFNN) excels in this scope. Despite needing considerably higher training time when compared to the other attempted models, it performs at the highest accuracy, with 0.48 root mean squared error (RMSE) and 99.72% Pearson coefficient for goodness of fit (R-squared). In contrast, the rigid regression method had the worst predictions with 4.92 RMSE and 37.29% R-squared. Also, random forest, boosting methods, and simple feed forward neural network can be considered as a middle accuracy model with faster training time than CFNN. The findings of this study not only advance modeling of superconductors but also pave the way for applications and further research on ML plug-and-play codes for superconducting studies including modeling of superconducting devices.

Design and Optimization of Power Shift Tractor Starting Control Strategy Based on PSO-ELM Algorithm

Power shift tractors have been widely used in agricultural tractors in recent years because of their advantages of uninterrupted power during shifting, high transmission efficiency and high stability. As one of the indispensable driving states of the power shift tractor, the starting process requires a small impact and a starting speed that meets the driver’s requirements. In this paper, aiming at such contradictory requirements, the starting control strategy of a power shift tractor is formulated with the goal of starting quality and the driver’s intention. Firstly, the identification characteristics of the driver under three starting intentions are obtained by a real vehicle test. An extreme learning machine with fast identification speed and short training time is used to establish the basic driver’s intention identification model. For the instability of the identification results of the Extreme Learning Machine (ELM), the particle swarm optimization algorithm (PSO) is used to optimize the ELM. The optimized extreme learning machine model has an accuracy of 96.891% for driver’s intention identification. The wet clutch is an important part of the power shift gearbox. In this paper, the starting control strategy knowledge base of the starting clutch is established by a combination of bench tests and simulation tests. Through the fuzzy algorithm, the driver’s intention is combined with the starting control strategy. Different drivers’ intentions will affect the comprehensive evaluation model of the clutch (the single evaluation index of the clutch is: the maximum sliding power, the sliding power, the speed stability time, the impact degree), thus affecting the final choice of the starting clutch control strategy considering the driver’s intention. On this basis, this paper studies and establishes the MPC starting controller for the power shift gearbox. Compared with the linear control strategy, the PSO-ELM-fuzzy weight starting strategy proposed in this paper can reduce the maximum sliding friction power by 45%, the sliding friction power by 69.45%, and the speed stabilization time by 0.11 s. The effectiveness of the starting control strategy considering the driver’s intention proposed in this paper to improve the starting quality of the power shift tractor is verified.

Diffusion models for conditional generation of hypothetical new families of superconductors

Effective computational search holds great potential for aiding the discovery of high-temperature superconductors (HTSs), especially given the lack of systematic methods for their discovery. Recent progress has been made in this area with machine learning, especially with deep generative models, which have been able to outperform traditional manual searches at predicting new superconductors within existing superconductor families but have yet to be able to generate completely new families of superconductors. We address this limitation by implementing conditioning—a method to control the generation process—for our generative model and develop SuperDiff, a denoising diffusion probabilistic model with iterative latent variable refinement conditioning for HTS discovery—the first deep generative model for superconductor discovery with conditioning on reference compounds. With SuperDiff, by being able to control the generation process, we were able to computationally generate completely new families of hypothetical superconductors for the very first time. Given that SuperDiff also has relatively fast training and inference times, it has the potential to be a very powerful tool for accelerating the discovery of new superconductors and enhancing our understanding of them.

Enhancements in Immediate Speech Emotion Detection: Harnessing Prosodic and Spectral Characteristics

Speech is essential to human communication for expressing and understanding feelings. Emotional speech processing has challenges with expert data sampling, dataset organization, and computational complexity in large-scale analysis. This study aims to reduce data redundancy and high dimensionality by introducing a new speech emotion recognition system. The system employs Diffusion Map to reduce dimensionality and includes Decision Trees and K-Nearest Neighbors(KNN)ensemble classifiers. These strategies are suggested to increase voice emotion recognition accuracy. Speech emotion recognition is gaining popularity in affective computing for usage in medical, industry, and academics. This project aims to provide an efficient and robust real-time emotion identification framework. In order to identify emotions using supervised machine learning models, this work makes use of paralinguistic factors such as intensity, pitch, and MFCC. In order to classify data, experimental analysis integrates prosodic and spectral information utilizing methods like Random Forest, Multilayer Perceptron, SVM, KNN, and Gaussian Naïve Bayes. Fast training times make these machine learning models excellent for real-time applications. SVM and MLP have the highest accuracy at 70.86% and 79.52%, respectively. Comparisons to benchmarks show significant improvements over earlier models.

When decoupled GCN meets group discrimination: A special graph contrastive learning framework

While Graph Contrastive Learning (GCL) has demonstrated satisfactory outcomes, three main problems remain. First, to effectively transmit the message of higher-order hop nodes, it is imperative to stack multiple graph convolution layers, which results in heightened computational demands and memory requirements. Second, the traditional GCL framework employs the pairwise comparison of positive and negative samples, which could slow model training and necessitate considerable computing resources. Third, the GCL paradigm places considerable emphasis on the data augmentation. Some augmentation techniques focus on altering the graph’s structure, which demand significant computational resources. In order to resolve the aforementioned problems in GCL, we incorporate the decoupled GCN into GCL paradigm and develop a new GCL framework termed Decoupled Group Discrimination (DGD), which relies on a one-layer Multi-Layer Perceptron (MLP) encoder and entirely abandons GCN. We also put forward a new negative sample generation strategy which is customized for DGD and enriches the graph structure information during training. Furthermore, we identify a more effective approach to aggregate information from individual nodes by a range of learnable parameters. Experiments on various datasets for the node classification reveal that DGD yields the superior accuracy, faster training time, and less memory consumption. Specifically, DGD improves the overall accuracy of the node classification by a margin of 1 to 2 percentages with the time required for the model training diminished by over 90% and the memory reduced exceeding 70%.

Ensemble feature selection using weighted concatenated voting for text classification

Following the increasing number of high dimensional data, selecting relevant features has always been better handled by filter feature selection techniques due to its improved generalization, faster training time, dimensionality reduction, less prone to overfitting, and improved model performance. However, the most used feature selection methods are unstable; a feature selection method chooses different subsets of characteristics that produce different classification accuracy. Selecting an appropriate hybrid harnesses the local feature relevant to the discriminative power of filter methods for improved text classification, which is lacking in past literature. In this paper, we proposed a novel multi-univariate hybrid feature selection method (MUNIFES) for enhanced discriminative power between the features and the target class. The proposed method utilizes multi-iterative processes to select the best feature sets from each univariate feature selection method. MUNIFES has employed the ensemble of multi-filter discriminative strength of Chi-Square (Chi2), Analysis of Variance (ANOVA), and Infogain methods to select optimal feature subsets. To evaluate the success of the proposed method, several experiments were performed on the 20newsgroup dataset and its variant (17newsgroup) with 10 classifiers (including ensemble, classification and optimization algorithms, and Artificial Neural Network (ANN)), compared with the state-of-the-art feature selection methods. The MUNIFES results indicated a better accuracy classification performance.

Comparison of Gradient Boosting and Random Forest Models in the Detection System of Rakaat during Prayer

Abstract. Errors in the execution of prayer among Muslims can occur due to a lack of profound understanding of the prayer procedure. This research aims to compare two machine learning models, Random Forest and Gradient Boosting, in classifying prayer movements, subsequently extending to calculate the number of prayer cycles (rakaat). A total of 7220 manually gathered data based on 33 landmark coordinates using Mediapipe Pose Detection were employed. The research findings reveal that the Random Forest model with a 70:30 ratio achieves 99.9% accuracy, precision, and recall, with the fastest training time being 3.8 seconds. Both models exhibit testing results close to 100%, but the Gradient Boosting model faces challenges in classifying specific movements. On the other hand, Random Forest successfully overcomes thesechallenges, enabling accurate prayer cycle calculations. The findings can contribute to the development of tools supporting Muslims in correct prayer execution, positively impacting religious and well-being aspects.

A multivariate method of forecasting the nonlinear dynamics of production network based on multilayer neural models

Design of production network based on multilayer neural models is considered in this paper. Design of production network is crucial because it determines the optimal location of production and logistics facilities, affects cost efficiency, customer service level and overall competitiveness in the global market. Multi-layer neural networks play an important role in this process, using advanced algorithms, machine learning models and optimization techniques to analyze huge amounts of data. Special attention is focused on qualitative analysis of dynamic behavior, dynamic lattice model. The model includes rate constants and initial conditions affecting the model trajectories, which can be classified as a stable site, limit cycle, or chaotic attractor. We aim to solve the problem of qualitative behavior of the model as a problem of multilayer neural models. A multivariate method of predicting nonlinear dynamics was used to construct the training data set. Neural networks defined by regenerative architectures with linear and non-linear outputs were analyzed and compared. As a result of the analysis, it was found that architectures with linear outputs show better correspondence between expected and predicted values. Architectures with non-linear outputs, despite their complexity, exhibit less accuracy and more deviation compared to linear ones. The single-layer architecture with linear outputs shows the best accuracy, although the two-layer architecture with linear outputs has the lowest rms error. Architectures with non-linear outputs have faster training times but poor accuracy, while architectures with linear outputs require more training time but have lower errors. The results obtained in the work indicate the importance of choosing the right architecture of the neural network depending on the tasks and requirements for accuracy and training time of the model.

DSE-NN: Deeply Supervised Efficient Neural Network for Real-Time Remote Photoplethysmography.

Non-contact remote photoplethysmography can be used in a variety of medical and healthcare fields by measuring vital signs continuously and unobtrusively. Recently, end-to-end deep learning methods have been proposed to replace the existing handcrafted features. However, since the existing deep learning methods are known as black box models, the problem of interpretability has been raised, and the same problem exists in the remote photoplethysmography (rPPG) network. In this study, we propose a method to visualize temporal and spectral representations for hidden layers, deeply supervise the spectral representation of intermediate layers through the depth of networks and optimize it for a lightweight model. The optimized network improves performance and enables fast training and inference times. The proposed spectral deep supervision helps to achieve not only high performance but also fast convergence speed through the regularization of the intermediate layers. The effect of the proposed methods was confirmed through a thorough ablation study on public datasets. As a result, similar or outperforming results were obtained in comparison to state-of-the-art models. In particular, our model achieved an RMSE of 1 bpm on the PURE dataset, demonstrating its high accuracy. Moreover, it excelled on the V4V dataset with an impressive RMSE of 6.65 bpm, outperforming other methods. We observe that our model began converging from the very first epoch, a significant improvement over other models in terms of learning efficiency. Our approach is expected to be generally applicable to models that learn spectral domain information as well as to the applications of regression that require the representations of periodicity.

MLGPnet: Multi-granularity neural network for 3D shape recognition using pyramid data

This paper presents a Multi-Granularity 3D shape recognition network comprising point-granularity, line-granularity, and Pyramid-granularity networks, as well as multi-granularity convolutional layers (MLGPnet). The network takes pyramid data with high-level features generated from mesh data as input. The point-granularity, line-granularity, and pyramid-granularity networks respectively generate features at the point, line, and pyramid levels. Finally, two multi-granularity convolutional layers merge the features from these different levels to generate more efficient 3D shape global features. Compared to some classical 3D shape recognition network models, the proposed network achieves superior results on three publicly general-purpose datasets. Notably, among all mesh-based recognition networks, the proposed network demonstrates the best recognition accuracy and retrieval rate. Furthermore, the proposed network model performs better in terms of training time and model complexity, with faster training time and fewer model parameters, resulting in faster recognition speed.

Reconstructing higher-resolution four-dimensional time-varying volumetric data

We have witnessed substantial growth in super-resolution research within the computer vision community. Unlike previous works that mainly focus on the super-resolution synthesis of images, videos, or single volumes, our research is dedicated to the super-resolution synthesis of time-varying volumetric data, which are generated from scientific simulations and are crucial for domain scientists to understand and analyse complex scientific phenomena. Compared to previous works, our research presents a greater challenge: the time-varying volumetric data have higher dimensions, making it more difficult to synthesise super-resolution that maintains good spatio-temporal consistency while achieving high visual quality. To tackle this challenge, we introduce a new GAN-based network called SSR-DoubleUNetGAN, which includes novel network architecture and loss functions, allowing for accurate synthesis of spatial super-resolution for time-varying volumetric data with relatively fast training time. Our method can be applied in the context of in-situ visualisation to aid domain scientists in analysing more time-varying volumetric data more efficiently. In addition, it can be used in the compression-decompression pipeline to recover the super-resolution time-varying volumetric data from their low-resolution counterpart. To demonstrate its effectiveness, we applied various time-varying volumetric datasets from different scientific simulations to it. To demonstrate its advantages, we compared it qualitatively and quantitatively with five state-of-the-art super-resolution techniques, namely SSR-TVD, Tricubic, SRResNet, Cubic, and Linear. Furthermore, we conducted an ablation study to validate its important modules. The experimental results show that our method outperforms the compared state-of-the-art techniques.

Lip Reading Using Various Deep Learning Models with Visual Turkish Data

In the Human-Computer Interaction field, lip reading is essential and still an open research problem. In the last decades, there have been many studies in the field of Automatic Lip-Reading (ALR) in different languages which is important for societies where the essential applications developed. Similarly to other machine learning and artificial intelligence applications, Deep Learning (DL) based classification algorithms have been applied for ALR in order to improve the performance of ALR. In the field of ALR, few studies have been done on the Turkish language. In this study, firstly an original data set was provided. Also, three image data augmentation techniques, which are sigmoidal transform, horizontal flip, and inverse transform, were applied to increase the data quality and variety. Then three deep learning models: Convolutional Neural Networks (CNN), Long-Short Term Memory (LSTM), and Bidirectional Gated Recurrent Unit (BGRU), were performed with a visual Turkish lip reading dataset. The performance of the applied method has been compared regarding precision, recall, and F1 metrics. According to experiment results, BGRU and LSTM models gave the same results up to the fifth decimal, and BGRU had the fastest training time.