Data Generation Algorithm Research Articles

Sensitivity-guided iterative parameter identification and data generation with BayesFlow and PELS-VAE for model calibration

Calibration of complex system models with a large number of parameters using standard optimization methods is often extremely time-consuming and not fully automated due to the reliance on all-inclusive expert knowledge. We propose a sensitivity-guided iterative parameter identification and data generation algorithm. The sensitivity analysis replaces manual intervention, the parameter identification is realized by BayesFlow allowing for uncertainty quantification, and the data generation with the physics-enhanced latent space variational autoencoder (PELS-VAE) between two iteration steps enables inference of weakly identifiable parameters. A complete calibration experiment was conducted on a thermal model of an automotive cabin. The average relative error rate over all parameter estimates of 1.62% and the mean absolute error of calibrated model outputs of 0.108,^{circ }C validate the feasibility and effectiveness of the method. Moreover, the entire procedure accelerates up to 1 day, whereas the classical calibration method takes more than 1 week.

Synthetic Patient Data Generation and Evaluation in Disease Prediction Using Small and Imbalanced Datasets

The increasing prevalence of chronic non-communicable diseases makes it a priority to develop tools for enhancing their management. On this matter, Artificial Intelligence algorithms have proven to be successful in early diagnosis, prediction and analysis in the medical field. Nonetheless, two main issues arise when dealing with medical data: lack of high-fidelity datasets and maintenance of patient's privacy. To face these problems, different techniques of synthetic data generation have emerged as a possible solution. In this work, a framework based on synthetic data generation algorithms was developed. Eight medical datasets containing tabular data were used to test this framework. Three different statistical metrics were used to analyze the preservation of synthetic data integrity and six different synthetic data generation sizes were tested. Besides, the generated synthetic datasets were used to train four different supervised Machine Learning classifiers alone, and also combined with the real data. F1-score was used to evaluate classification performance. The main goal of this work is to assess the feasibility of the use of synthetic data generation in medical data in two ways: preservation of data integrity and maintenance of classification performance.

Privacy-utility equilibrium data generation based on Wasserstein generative adversarial networks

To solve the contradictory problem of local data sharing and privacy protection, this paper presented the models and algorithms for privacy-utility equilibrium data generation based on Wasserstein generative adversarial networks. First, we introduce the expected estimation error between the discriminant probability of real data and generated data into Wasserstein generative adversarial networks, and we formally construct a basic mathematical model of privacy-utility equilibrium data generation based on computationally indistinguishable. Second, we construct the basic model and the basic algorithm of privacy-utility equilibrium data generation based on Wasserstein generative adversarial networks according to the constructed basic mathematical model, and our theoretical analysis results show that the basic algorithm can achieve the equilibrium between local data sharing and privacy protection. Third, according to the constructed basic model, we construct the federated model and the federated algorithm of privacy-utility equilibrium data generation based on Wasserstein generative adversarial networks using the serialized training method of federated learning, and our theoretical analysis results also show that the federated algorithm can achieve the equilibrium between local data sharing and privacy protection in a distributed environment. Finally, our experimental results show that the proposed algorithms can achieve the equilibrium between local data sharing and privacy protection in this paper. Therefore, the constructed basic mathematical model provides a theoretical basis of achieving the equilibrium between local data sharing and privacy protection. At the same time, the proposed basic model and the federated model of privacy-utility equilibrium data generation provide a concrete method of achieving the equilibrium between local data sharing and privacy protection in centralized and distributed environment respectively.

ПРИМЕНЕНИЕ ГЛУБОКОГО ОБУЧЕНИЯ ДЛЯ АУГМЕНТАЦИИ И ГЕНЕРАЦИИ ПОДВОДНОГО НАБОРА ДАННЫХ С ПРОМЫШЛЕННЫМИ ОБЪЕКТАМИ

The purpose of the study: development of a deep learning method for augmentation and generation of a problem-oriented dataset containing industrial objects, including the development of more efficient data generation and augmentation algorithms based on deep learning, which allow you to create more diverse and realistic data corresponding to industrial objects that can be transferred from one style domain to another. The goal set in the study is related to the actual scientific and technical problem of providing computer vision in systems operating in the underwater environment. These can be autonomous uninhabited underwater vehicles that look for breaks in pipelines, analyze oil leaks, the movement of schools of fish, etc. However, today there is not enough data containing the described objects in the conditions of their real existence. Thus, it is necessary to provide the training sample with realistic images. Research methods: the CycleGAN architecture, which converts a dataset containing images of various objects taken in a laboratory or in a conventional aboveground environment into a dataset containing the cha¬racteristics of an underwater environment. To evaluate the developed augmentation algorithm, it is proposed to use image classification by domains, which can be performed using the ResNet convolutional neural network. Results of the study. A tool is presented to solve the problem of the lack of underwater datasets, a deep learning model is developed, which is used to create images with underwater elements. The model works on the principle of a cyclic generative adversarial network, which receives a real image of an industrial facility in surface conditions as an input, and returns a generated image of the same industrial facility in underwater conditions as an output. Visual analysis of images shows that this method is quite adequate. In addition, a test on the classification model showed almost 100% ability of the neural network to distinguish between domains. Conclusion. The study showed that the CycleGAN model can be used to create images of various objects in the underwater environment. In the future, it is possible to search for additional augmentation procedures, in addition, augmentations of the generated set can be used images, which will also provide researchers and developers with sufficient material with industrial facilities in the underwater environment. This can improve the quality of developments.

Skill-level classification and performance evaluation for endoscopic sleeve gastroplasty.

We previously developed grading metrics for quantitative performance measurement for simulated endoscopic sleeve gastroplasty (ESG) to create a scalar reference to classify subjects into experts and novices. In this work, we used synthetic data generation and expanded our skill level analysis using machine learning techniques. We used the synthetic data generation algorithm SMOTE to expand and balance our dataset of seven actual simulated ESG procedures using synthetic data. We performed optimization to seek optimum metrics to classify experts and novices by identifying the most critical and distinctive sub-tasks. We used support vector machine (SVM), AdaBoost, K-nearest neighbors (KNN) Kernel Fisher discriminant analysis (KFDA), random forest, and decision tree classifiers to classify surgeons as experts or novices after grading. Furthermore, we used an optimization model to create weights for each task and separate the clusters by maximizing the distance between the expert and novice scores. We split our dataset into a training set of 15 samples and a testing dataset of five samples. We put this dataset through six classifiers, SVM, KFDA, AdaBoost, KNN, random forest, and decision tree, resulting in 0.94, 0.94, 1.00, 1.00, 1.00, and 1.00 accuracy, respectively, for training and 1.00 accuracy for the testing results for SVM and AdaBoost. Our optimization model maximized the distance between the expert and novice groups from 2 to 53.72. This paper shows that feature reduction, in combination with classification algorithms such as SVM and KNN, can be used in tandem to classify endoscopists as experts or novices based on their results recorded using our grading metrics. Furthermore, this work introduces a non-linear constraint optimization to separate the two clusters and find the most important tasks using weights.

An Offline Weighted-Bagging Data-Driven Evolutionary Algorithm with Data Generation Based on Clustering

In recent years, a variety of data-driven evolutionary algorithms (DDEAs) have been proposed to solve time-consuming and computationally intensive optimization problems. DDEAs are usually divided into offline DDEAs and online DDEAs, with offline DDEAs being the most widely studied and proven to display excellent performance. However, most offline DDEAs suffer from three disadvantages. First, they require many surrogates to build a relatively accurate model, which is a process that is redundant and time-consuming. Second, when the available fitness evaluations are insufficient, their performance tends to be not entirely satisfactory. Finally, to cope with the second problem, many algorithms use data generation methods, which significantly increases the algorithm runtime. To overcome these problems, we propose a brand-new DDEA with radial basis function networks as its surrogates. First, we invented a fast data generation algorithm based on clustering to enlarge the dataset and reduce fitting errors. Then, we trained radial basis function networks and carried out adaptive design for their parameters. We then aggregated radial basis function networks using a unique model management framework and demonstrated its accuracy and stability. Finally, fitness evaluations were obtained and used for optimization. Through numerical experiments and comparisons with other algorithms, this algorithm has been proven to be an excellent DDEA that suits data optimization problems.

Three-stage data generation algorithm for multiclass network intrusion detection with highly imbalanced dataset

The Internet plays a crucial role in our daily routines. Ensuring cybersecurity to Internet users will provide a safe online environment. Automatic network intrusion detection (NID) using machine learning algorithms has recently received increased attention recently. The NID model is prone to bias towards the classes with more training samples due to highly imbalanced datasets across different types of attacks. The challenge in generating additional training data for minority classes is the generation of insufficient data. The study's purpose is to address this challenge, which extends the data generation ability by proposing a three-stage data generation algorithm using the synthetic minority over-sampling technique, a generative adversarial network (GAN), and a variational autoencoder. A convolutional neural network is employed to extract the representative features from the data, which were fed into a support vector machine with a customised kernel function. An ablation study evaluated the effectiveness of the three-stage data generation, feature extraction, and customised kernel. This was followed by a performance comparison between our study and existing studies. The findings revealed that the proposed NID model achieved an accuracy of 91.9%–96.2% in the four benchmark datasets. In addition, it outperformed existing methods such as GAN-based deep neural networks, conditional Wasserstein GAN-based stacked autoencoder, synthesised minority oversampling technique-based random forest, and variational autoencoder-based deep neural network, by 1.51%–28.4%.

NIRS Data Augmentation Technique to Detect Hemodynamic Peaks During Self-Paced Motor Imagery

Optical brain monitoring, such as near-infrared spectroscopy (NIRS), has facilitated numerous brain studies, including those based on machine learning techniques. A large and diverse dataset is necessary for training machine learning algorithms to avoid overfitting a limited amount of data. However, recruiting sufficient subjects is challenging owing to time and budget constraints. Therefore, we propose an NIRS data generation algorithm that scales NIRS signal components, such as hemodynamic response function, physiological noise, and system spike noise, based on the source-detector distance to augment the training data. The experimental data were augmented with the generated NIRS data to train a convolutional neural network to classify self-paced left- and right-hand motor imagery. Augmenting the training dataset with 1000 generated data points increased the classification accuracy to 86.3 ± 4.1%, indicating a 26% increase compared with training on experimental data only. In addition, we applied Guided Gradient-weighted Class Activation Mapping (Grad-CAM) to visualize the class discriminative features of the input data. The Guided Grad-CAM heatmaps aligned well with the oxy-hemoglobin peaks during self-paced motor imagery. We concluded that the increased cerebral oxygenation, especially in the contralateral hemisphere, was the class-discriminative feature for classifying left- and right-hand motor imagery.

Automated Cognitive Health Assessment Using Partially Complete Time Series Sensor Data.

Behavior and health are inextricably linked. As a result, continuous wearable sensor data offer the potential to predict clinical measures. However, interruptions in the data collection occur, which create a need for strategic data imputation. The objective of this work is to adapt a data generation algorithm to impute multivariate time series data. This will allow us to create digital behavior markers that can predict clinical health measures. We created a bidirectional time series generative adversarial network to impute missing sensor readings. Values are imputed based on relationships between multiple fields and multiple points in time, for single time points or larger time gaps. From the complete data, digital behavior markers are extracted and are mapped to predicted clinical measures. We validate our approach using continuous smartwatch data for n = 14 participants. When reconstructing omitted data, we observe an average normalized mean absolute error of 0.0197. We then create machine learning models to predict clinical measures from the reconstructed, complete data with correlations ranging from r = 0.1230 to r = 0.7623. This work indicates that wearable sensor data collected in the wild can be used to offer insights on a person's health in natural settings.

Leveraging deep learning algorithms for synthetic data generation to design and analyze biological networks

The use of synthetic data is gaining an increasingly prominent role in data and machine learning workflows to build better models and conduct analyses with greater statistical inference. In the domains of healthcare and biomedical research, synthetic data may be seen in structured and unstructured formats. Concomitant with the adoption of synthetic data, a sub-discipline of machine learning known as deep learning has taken the world by storm. At a larger scale, deep learning methods tend to outperform traditional methods in regression and classification tasks. These techniques are also used in generative modeling and are thus prime candidates for generating synthetic data in both structured and unstructured formats. Here, we emphasize the generation of synthetic data in healthcare and biomedical research using deep learning methods for unstructured data formats such as text and images. Deep learning methods leverage the neural network algorithm, and in the context of generative modeling, several neural network architectures can create new synthetic data for a problem at hand including, but not limited to, recurrent neural networks (RNNs), variational autoencoders (VAEs), and generative adversarial networks (GANs). To better understand these methods, we will look at specific case studies such as generating realistic clinical notes of a patient, the generation of synthetic DNA sequences, as well as to enrich experimental data collected during the study of heterotypic cultures of cancer cells.

Cardinality estimation of approximate substring queries using deep learning

Cardinality estimation of an approximate substring query is an important problem in database systems. Traditional approaches build a summary from the text data and estimate the cardinality using the summary with some statistical assumptions. Since deep learning models can learn underlying complex data patterns effectively, they have been successfully applied and shown to outperform traditional methods for cardinality estimations of queries in database systems. However, since they are not yet applied to approximate substring queries, we investigate a deep learning approach for cardinality estimation of such queries. Although the accuracy of deep learning models tends to improve as the train data size increases, producing a large train data is computationally expensive for cardinality estimation of approximate substring queries. Thus, we develop efficient train data generation algorithms by avoiding unnecessary computations and sharing common computations. We also propose a deep learning model as well as a novel learning method to quickly obtain an accurate deep learning-based estimator. Extensive experiments confirm the superiority of our data generation algorithms and deep learning model with the novel learning method.

인공지능(AI)을 활용한 작곡 기술 동향과 창작 플랫폼

this study, the technical concept of composition using artificial intelligence, which is attracting attention as a major field of the 4th industrial revolution, and the development and service cases of major platforms are studied. The purpose of this study was to find out the plan and development plan. The concept of artificial intelligence was explained, and the concepts of LSTM and RNN, which are representative data generation models used in artificial intelligence composition, and GAN, which is an adversarial generation model, were studied, and the data generation process and algorithm of artificial intelligence were investigated. In addition, by analyzing the case of a music creation platform using artificial intelligence, which is a representative use case of the current artificial intelligence composition, the status of technology and service was investigated, and the creation process using the artificial intelligence composition platform was investigated. In addition, by analyzing the case of a music creation platform using artificial intelligence, which is a representative use case of the current artificial intelligence composition, the status of technology and service was investigated, and the creation process using the artificial intelligence composition platform was investigated.

Analysis of methods and algorithms for generating key data for FALCON-like electronic signature algorithms

At present and in the future, mathematical methods, mechanisms and algorithms of standardized asymmetric cryptotransformations such as electronic signature (ES) are and will be used for information cryptographic protection. Electronic signature is the main and essential component of cybersecurity, in terms of providing quality information security services such as integrity, irresistibility and authenticity of information and data being processed. However, there are well-founded suspicions that in the post-quantum period the existing ES standards will be broken and compromised using classical and quantum cryptanalytic systems with appropriate mathematical, software and hardware-software. An analysis was performed, which confirms that quantum computers have already been developed, manufactured and used. This work is devoted to the analysis of methods and algorithms for generating key data for Falcon-like algorithms of electronic signature. Some of the basic algorithms for Falcon-shaped algorithms of electronic signature are considered, namely the algorithm of key data generation and algorithm of random polynomials f, g generation, which satisfy the Gauss distribution. The Falcon algorithm itself is the finalist of the post-quantum electronic signature contest due to the satisfactory value of the public key and signature lengths, but the key data generation algorithm uses many methods and difficult to implement. The Falcon authors use this algorithm for polynomials n=512, 1024. To increase the sixth level of cryptostability, this algorithm can be expanded for n=2048. This work is devoted to study the Falcon algorithm, taking into account its expansion for n=512, 1024, 2048 in terms of generating key data. Also, the paper considers the results of justifying the choice of a mathematical apparatus for implementing a software package for generating a key pair of a cryptographic algorithm for an electronic signature in order to create reliable electronic signatures.

Integration of machine learning with economic energy scheduling

The aim of economic load dispatch (ELD) is to deliver required electrical power for a specified period at the lowest possible generation cost using available generating units (GUs). It is imperative to lower the generation costs in order to reduce the consumer costs and to generate adequate revenue from large capital investments in the power sector. There are several optimization algorithms (OAs) to solve this issue. In this study, a new method that combines machine learning (ML) with an OA is used to come up with a high-precision, best solution for ELD issues in the quickest time possible. The ’Lagrange Multiplier’ (LM) method is used as the OA, while the ’Decision Tree’ (DT) algorithm is used as the ML algorithm. ML algorithms require data to train themselves. A data generation algorithm (DGA) is used to generate data considering constraints such as the power balance constraint, transmission loss (TL), generating capacity, and prohibited operating zones (POZs). The DGA is based on the LM method with constraint handling techniques. Without considering ramp rate limits (RRLs), the optimal load sharing data is generated over the whole power capacity range of the committed GUs. The power capacity ranges from the sum of the minimum power capacity to the maximum power capacity of the committed GUs. This range is divided into several discrete data points with a step size of 0.01. Optimal load sharing among the GUs has been calculated for each of the data points using DGA. Then the DT model was trained with the generated data that could have been used further to predict the load sharing among the GUs. To impose RRLs, we have developed a search method using the trained DT model. We have validated our proposed method through three case studies: Case 1: 6 GUs with a 1263 MW power demand; Case 2: 15 GUs with a 2630 MW power demand; and Case 3: 140 GUs with a 49342 MW power demand. Finally, the optimal solution for all the case studies using the proposed method was compared with the existing methods. The proposed method was found to be better than the existing methods in terms of time, precision, and cost. This opens up a new way to help with the ELD issue by combining ML with OA.

Autoencoder-guided GAN for minority-class cloth-changing gait data generation

In the gait recognition task, cloth-changing data (e.g., CL-class data in the CASIA-B dataset) account for a very low proportion of the gait dataset, which brings serious recognition accuracy degradation to gait recognition algorithm in clothes changing scenarios. One way to solve this problem is to use data generation algorithms to expand the cloth-changing data. When facing the generation of minority-class data, the existing GAN-based image data expansion algorithms require a large amount of data to ensure the convergence, while mode collapse can occur due to the conflict between the true-false discriminant loss and the classification loss. In order to solve the above problems, we propose an autoencoder-guided GAN for cloth-changing gait data generation, which uses the autoencoder with embedding layer to extract the data features of the entire data set, rather than only extracting the features of the required CL data. Then the labeled latent vector (representing the extracted data features) in the autoencoder is taken as the input of the GAN, solving the problem that GAN convergence requires a lot of training data. The GAN has the same topology as the autoencoder and uses the parameter matrix of the autoencoder as the initial weights. Secondly, the true classification in GAN is canceled, and only the false classification is retained, which solves the problem of the collision of the loss function between the true-false discrimination and the classification discrimination. In the cloth-changing data generation experiments, compared with other GAN data expansion algorithms, our method generates clothes changing data with high quality, and ensures the diversity of the generated data. When using the CASIA-B dataset expanded by our algorithm, several gait recognition algorithms all show significant improvement, especially in the clothes changing scenario.

CCR-GSVM: A boundary data generation algorithm for support vector machine in imbalanced majority noise problem

CCR-GSVM: A boundary data generation algorithm for support vector machine in imbalanced majority noise problem

CUDA Implementation For Eye Location On Infrared Images

Parallel programming using GPUs is a modern solution to reduce computation time for large tasks. This is done by dividing algorithms in smaller parts which can be executed simultaneously. CUDA has many practical applications especially in video processing, medical imaging and machine learning. This paper presents how parallel implementations can speedup a ground truth data generation algorithm for eye location on infrared driver recordings which is executed on a database with more than 2 million frames. Computation time is much shorter compared to a sequential CPU implementation which makes it feasible to run it multiple times if updates are required and even use it in real time applications.

LayeredCNN: Segmenting Layers with Autoregressive Models

We address a subclass of segmentation problems where the labels of the image are structured in layers. We propose applying autoregressive CNNs which, when given an image and a partial segmentation of layers, complete the segmentation. Initializing the model with a user-provided partial segmentation allows for choosing which layers the model should segment. Alternatively, the model can produce an automatic initialization, albeit with some performance loss. The model is trained exclusively on synthetic data from our data generation algorithm. It yields impressive performance on the synthetic data and generalizes to real data it has never seen.

Prediction of Preterm Delivery from Unbalanced EHG Database.

Objective: The early prediction of preterm labor can significantly minimize premature delivery complications for both the mother and infant. The aim of this research is to propose an automatic algorithm for the prediction of preterm labor using a single electrohysterogram (EHG) signal. Method: The proposed method firstly employs empirical mode decomposition (EMD) to split the EHG signal into two intrinsic mode functions (IMFs), then extracts sample entropy (SampEn), the root mean square (RMS), and the mean Teager–Kaiser energy (MTKE) from each IMF to form the feature vector. Finally, the extracted features are fed to a k-nearest neighbors (kNN), support vector machine (SVM), and decision tree (DT) classifiers to predict whether the recorded EHG signal refers to the preterm case. Main results: The studied database consists of 262 term and 38 preterm delivery pregnancies, each with three EHG channels, recorded for 30 min. The SVM with a polynomial kernel achieved the best result, with an average sensitivity of 99.5%, a specificity of 99.7%, and an accuracy of 99.7%. This was followed by DT, with a mean sensitivity of 100%, a specificity of 98.4%, and an accuracy of 98.7%. Significance: The main superiority of the proposed method over the state-of-the-art algorithms that studied the same database is the use of only a single EHG channel without using either synthetic data generation or feature ranking algorithms.

Data augmentation of credit default swap transactions based on a sequence GAN

Credit default swap transaction data repositories are frequently applied with credit default swap spread estimation and financial market risk assessment. However, in practical applications, there is poor liquidity, some missing data, and inaccurate definitions. Small samples tend to lead to poor prediction accuracy and poor adaptability of the statistical algorithm. Data generation can effectively increase the sample size and improve the effect of the risk assessment model. In this paper, a credit default swap data generation algorithm based on a sequence generative adversarial network (SeqGAN) is proposed, and the policy gradient algorithm in reinforcement learning is introduced to optimize the traditional generative adversarial network (GAN) algorithm to solve the gradient disappearance and poor data adaptability problems in the traditional algorithm. Gradient disappearance is due to the generator network in GAN being designed to be able to adjust the output continuously, which does not work on discrete data generation. The optimization algorithm proposed in this paper is used to train randomly distributed sequence data and generate credit default swap transactions with diversity and good model applicability. The credit default swap data generated in this paper are verified by the synthetic ranking agreement (SRA) index. The results show that SeqGAN can effectively synthesize various simulation samples, which can provide support for the risk discrimination model.