Huge Training Data Research Articles

Sharing massive biomedical data at magnitudes lower bandwidth using implicit neural function

Efficient storage and sharing of massive biomedical data would open up their wide accessibility to different institutions and disciplines. However, compressors tailored for natural photos/videos are rapidly limited for biomedical data, while emerging deep learning-based methods demand huge training data and are difficult to generalize. Here, we propose to conduct Biomedical data compRession with Implicit nEural Function (BRIEF) by representing the target data with compact neural networks, which are data specific and thus have no generalization issues. Benefiting from the strong representation capability of implicit neural function, BRIEF achieves 2[Formula: see text]3 orders of magnitude compression on diverse biomedical data at significantly higher fidelity than existing techniques. Besides, BRIEF is of consistent performance across the whole data volume, and supports customized spatially varying fidelity. BRIEF's multifold advantageous features also serve reliable downstream tasks at low bandwidth. Our approach will facilitate low-bandwidth data sharing and promote collaboration and progress in the biomedical field.

Graph Neural Network-based surrogate model for granular flows

Accurate simulation of granular flow dynamics is crucial for assessing geotechnical risks, including landslides and debris flows. Traditional numerical methods are limited by their computational cost in simulating large-scale systems. Statistical or machine learning-based models offer alternatives. Still, they are largely empirical, based on limited parameters. Due to their permutation-dependent learning, traditional machine learning-based models require huge training data to generalize. To resolve these problems, we use a graph neural network (GNN), a state-of-the-art machine learning architecture that learns local interactions. Graphs represent the state of dynamically changing granular flows and their interactions. We implement a multi-Graphics Processing Units (GPU) GNN simulator (GNS) capable of handling different material types. We demonstrate the capability of GNS by modeling granular flow interactions with barriers. GNS takes the granular flow’s current state and predicts the next state using Euler explicit integration by learning the local interaction laws. We train GNS on different granular trajectories. We then assess its performance by predicting granular column collapse and interaction with barriers. GNS accurately predicts flow dynamics for column collapses with different aspect ratios and interaction with barriers with configurations unseen during training. GNS is up to a few thousand times faster than high-fidelity numerical simulators.

Few‐shot learning for word‐level scene text script identification

AbstractScript identification of text in scene images has attracted massive attention recently. However, the existing techniques primarily emphasize on scripts where data are available abundantly, such as English, European, or East Asian. Although these methods are robust in dealing with high‐resource data, how these techniques will work on low‐resource scripts has yet to be discovered. For example, in India, there is a disparity among the text scripts across the country's demographic. To bridge the research gap for resource‐constraint script identification, we present a few‐shot learning network called the TextScriptFSLNet. This network does not require huge training data while achieving state‐of‐the‐art performance on benchmark datasets. Our proposed method acts in accordance with a ‐way ‐shot paradigm by splitting the train set as support and query set, respectively. The support set learns representative knowledge of each class and creates its prototypes. We use multi‐kernel spatial attention fused 2‐layer convolutional neural network and averaging technique to generate the prototype of each class. This spatial attention aids in grasping important information in an image and enriches the feature representation. To the best of our knowledge, the proposed work is the first of its kind in the scene text understanding domain. Additionally, we created a dataset called Indic‐FSL2023 comprising 10 of the 22 officially recognized Indian scripts. The proposed method achieves the highest accuracy among the tested methods on the newly created Indic‐FSL2023. Experiments are also conducted on MLe2e to demonstrate its versatility. Furthermore, we also showed how our proposed model performed concerning illumination changes and blur on scene text script images.

SiamDF: Tracking training data-free siamese tracker

Much progress has been made in siamese tracking, primarily benefiting from increasing huge training data. However, very little attention has been really paid to the role of huge training data in learning an effective siamese tracker. In this study, we undertake an in-depth analysis of this issue from a novel optimization perspective, and observe that training data is particularly adept at background suppression, thereby refining target representation. Inspired by this insight, we present a data-free siamese tracking algorithm named SiamDF, which requires only a pre-trained backbone and no further fine-tuning on additional training data. Particularly, to suppress background distractors, we separately improve two branches of siamese tracking by retaining the pure target region as target input with the removal of template background, and by exploring an efficient inverse transformation to maintain the constant aspect ratio of target state in search region. Besides, we further promote the center displacement prediction of the entire backbone by eliminating its spatial stride deviations caused by convolution-like quantification operations. Our experimental results on several popular benchmarks demonstrate that SiamDF, free from both offline fine-tuning and online update, achieves impressive performance compared to well-established unsupervised and supervised tracking methods.

Physics-Driven Synthetic Data Learning for Biomedical Magnetic Resonance: The imaging physics-based data synthesis paradigm for artificial intelligence

Deep learning (DL) has driven innovation in the field of computational imaging. One of its bottlenecks is unavailable or insufficient training data. This article reviews an emerging paradigm, imaging physics-based data synthesis (IPADS), that can provide huge training data in biomedical magnetic resonance (MR) without or with few real data. Following the physical law of MR, IPADS generates signals from differential equations or analytical solution models, making learning more scalable and explainable and better protecting privacy. Key components of IPADS learning, including signal generation models, basic DL network structures, enhanced data generation, and learning methods, are discussed. Great IPADS potential has been demonstrated by representative applications in fast imaging, ultrafast signal reconstruction, and accurate parameter quantification. Finally, open questions and future work are discussed.

CST-RL: Contrastive Spatio-Temporal Representations for Reinforcement Learning

Learning representations from high-dimensional observations is critical for training of pixel-based continuous control tasks with reinforcement learning (RL). Without proper representations, the training will be very inefficient, requiring long training time and huge training data to learn directly from low-level pixel observations. Yet, a lot of information in such observations may be redundant or irrelevant. A common approach to solving this problem is to train auxiliary objectives alongside the main RL objective. The additional objectives provide more signals to the model and reduce the training time, resulting in better sample efficiency. A representative work is Contrastive Unsupervised Representations for Reinforcement Learning (CURL), which leverages contrastive learning to assist RL to learn useful representations. Although CURL performs very well in extracting spatial information from pixel inputs, it is found to overlook potential temporal signals. In this paper, a contrastive spatio-temporal representation learning framework for RL, called CST-RL, is introduced, which leverages 3D Convolutional Neural Network (3D CNN) alongside contrastive learning for sample-efficient RL. It pays attention to both spatial and temporal signals in pixel observations. Experiments based on DMControl show that CST-RL outperforms CURL in all six environments after 500K environment steps and only needs half of the steps to achieve the standard score in the majority of cases.

Data augmentation by combining feature selection and color features for image classification

<span lang="EN-US">Image classification is an essential task in computer vision with various applications such as bio-medicine, industrial inspection. In some specific cases, a huge training data is required to have a better model. However, it is true that full label data is costly to obtain. Many basic pre-processing methods are applied for generating new images by translation, rotation, flipping, cropping, and adding noise. This could lead to degrade the performance. In this paper, we propose a method for data augmentation based on color features information combining with feature selection. This combination allows improving the classification accuracy. The proposed approach is evaluated on several texture datasets by using local binary patterns features.</span>

One-vs-All Convolutional Neural Networks for Synthetic Aperture Radar Target Recognition

Abstract Convolutional Neural Networks (CNN) have been widely utilized for Automatic Target Recognition (ATR) in Synthetic Aperture Radar (SAR) images. However, a large number of parameters and a huge training data requirements limit CNN’s use in SAR ATR. While previous works have primarily focused on model compression and structural modification of CNN, this paper employs the One-Vs-All (OVA) technique on CNN to address these issues. OVA-CNN comprises several Binary classifying CNNs (BCNNs) that act as an expert in correctly recognizing a single target. The BCNN that predicts the highest probability for a given target determines the class to which the target belongs. The evaluation of the model using various metrics on the Moving and Stationary Target Acquisition and Recognition (MSTAR) benchmark dataset illustrates that the OVA-CNN has fewer weight parameters and training sample requirements while exhibiting a high recognition rate.

Deep anomaly detection model for composite inspection in quadratic frequency modulated thermal wave imaging

Thermographic deep learning models are object dependent and require huge training data to customize the model. On the other hand, the thermographic data is class imbalanced and scarce. This article proposes a novel unsupervised, object independent automatic deep anomaly detection model through an experimental investigation on two composite specimens using quadratic frequency modulated thermal wave imaging. The proposed methodology employs transfer learning to extract the deep features from thermal profiles by using only sound region features and surpassing the cross-material inspection limitation of supervised deep neural network models and provides better performance as quantified using F-Score and AUC. A depth map based visualization is further presented using chirp z-transform phase-based depth estimation of anomalies to facilitate an easier characterization.

System identification method based on interpretable machine learning for unknown aircraft dynamics

Data-driven discovery of dynamics via machine learning has become the frontier work of modeling and control, and also provides great help for expanding the application range of model-based control methods. However, many machine learning methods need huge training data, and the generalization ability outside the training area is limited. These factors impede the development of machine learning in fields such as aerospace. To solve this problem, a new interpretable learning algorithm for aircraft systems is studied, which can consider the influence of control input and be implemented online to respond quickly to the system changes. Simulation results show that compared with the conventional neural network method, the proposed algorithm has higher performance, fewer data demands, and higher computational efficiency. Finally, the model identified online by the proposed algorithm is used in the model-based controller to further verify the effectiveness of this algorithm.

Chatbot integrated with machine learning deployed in the cloud and performance evaluation

Recently human-machine digital assistants gained popularity and commonly used in question-and-answer applications and similar consumer-supporting domains. A class of more sophisticated digital assistants employing longer dialogs follow the trend, and there are several commercial platforms supporting their prototyping such as Google DialogFlow, Manychat, Chatfuel, Amazon Lex, etc. This paper explores cloud deployment of chatbots systems and their performance assessment methodologies. The performance measures includes system response delays and natural language processing capabilities. A case study platform supporting so-called deep-logic chatbots with long cycling capabilities is implemented and used for the assessment. To enable human-like conversations with a chatbot, huge training data, complex natural language understanding models are required and need to be adjusted and trained continuously. We explore implementation formats supporting auto scaling, and uninterrupted availability. In particular, we employ an architecture consisting of separate dialog management, authentication, and Natural Language Understanding (NLU) services. Finally, we present a performance evaluation of such loosely coupled chatbot system. Keywords: Cloud Deployment, Natural language understanding, Chatbot, Performance assessment

Developing a robust and flexible smart tool to predict a full range Critical Heat Flux (CHF) in different LWRs by using deep learning Artificial Neural Networks (ANN) via parallel multi-processing

In this paper, a practical smart tool has been developed to predict CHF in LWRs. The tool is based on the strengthen cascade feed forward ANNs in addition of some of the deep learning features. The most complete experimental CHF data available for LWRs are chosen as the training and test data sets. Input layer can get different thermodynamic qualities, mass flow rates, and pressures to predict the CHF. Then it gains a full range CHF simulator for LWRs fortunately covering both anticipated and unknown transients. Although training such a huge ANN structure and the huge training data set is time consuming, but after the training it acts fast especially via a multi-processing platform. Results are quite satisfactory due to the very good network performance and regressions. Briefly, this is a fast, flexible, and robust tool to predict the CHF deep inside the physics of the most broadening CHF data set ever published for the LWRs.

Deep learning‐based optimal tracking control of flow front position in an injection molding machine

AbstractThe product quality of injection‐molded plastic is closely related to the injection flow velocity of molten plastics. In this article, an optimal tracking control problem for the injection flow front position arising in the filling process in the injection molding machine (IMM) is considered, and an intelligent real‐time optimal control method based on deep neural networks (DNNs) is developed for the online tracking of the flow front position to improve the efficient production process of the plastics. To this end, a highly nonlinear dynamic model describing the filling process is first proposed and then an optimal control problem is formulated. A high‐fidelity numerical optimization approach known as the Gauss pseudospectral method is used to numerically obtain the optimal state‐control solutions, which can then be saved as a huge training data set for the DNNs. The DNNs architecture is then developed and offline trained using the obtained data set to determine the best state–control relationship. As a consequence, the well‐trained DNN may be utilized to produce the matching optimum feedback action on‐board. Numerical studies are also carried out to verify the performance of our proposed approach.

Convolutional Neural Network with a Learnable Spatial Activation Function for SAR Image Despeckling and Forest Image Analysis

Synthetic aperture radar (SAR) images are often disturbed by speckle noise, making SAR image interpretation tasks more difficult. Therefore, speckle suppression becomes a pre-processing step. In recent years, approaches based on convolutional neural network (CNN) achieved good results in synthetic aperture radar (SAR) images despeckling. However, these CNN-based SAR images despeckling approaches usually require large computational resources, especially in the case of huge training data. In this paper, we proposed a SAR image despeckling method using a CNN platform with a new learnable spatial activation function, which required significantly fewer network parameters without incurring any degradation in performance over the state-of-the-art despeckling methods. Specifically, we redefined the rectified linear units (ReLU) function by adding a convolutional kernel to obtain the weight map of each pixel, making the activation function learnable. Meanwhile, we designed several experiments to demonstrate the advantages of our method. In total, 400 images from Google Earth comprising various scenes were selected as a training set in addition to 10 Google Earth images including athletic field, buildings, beach, and bridges as a test set, which achieved good despeckling effects in both visual and index results (peak signal to noise ratio (PSNR): 26.37 ± 2.68 and structural similarity index (SSIM): 0.83 ± 0.07 for different speckle noise levels). Extensive experiments were performed on synthetic and real SAR images to demonstrate the effectiveness of the proposed method, which proved to have a superior despeckling effect and higher ENL magnitudes than the existing methods. Our method was applied to coniferous forest, broad-leaved forest, and conifer broad-leaved mixed forest and proved to have a good despeckling effect (PSNR: 23.84 ± 1.09 and SSIM: 0.79 ± 0.02). Our method presents a robust framework inspired by the deep learning technology that realizes the speckle noise suppression for various remote sensing images.

A Payload Based Malicious HTTP Traffic Detection Method Using Transfer Semi-Supervised Learning

Malicious HTTP traffic detection plays an important role in web application security. Most existing work applies machine learning and deep learning techniques to build the malicious HTTP traffic detection model. However, they still suffer from the problems of huge training data collection cost and low cross-dataset generalization ability. Aiming at these problems, this paper proposes DeepPTSD, a deep learning method for payload based malicious HTTP traffic detection. First, it treats the malicious HTTP traffic detection as a text classification problem and trains the initial detection model using TextCNN on a public dataset, and then adapts the initial detection model to the target dataset based on a transfer learning algorithm. Second, in the transfer learning procedure, it uses a semi-supervised learning algorithm to accomplish the model adaptation task. The semi-supervised learning algorithm enhances the target dataset based on a HTTP payload data augmentation mechanism to exploit both the labeled and unlabeled data. We evaluate DeepPTSD on two real HTTP traffic datasets. The results show that DeepPTSD has competitive performance under the small data condition.

RETRACTED ARTICLE: Classification of audio signals using SVM-WOA in Hadoop map-reduce framework

Audio classification is the reason for the multimedia gratified examination which is the utmost significant and generally utilized application these days. For huge information bases, programmed classification procedure utilizing Artificial Intelligence (AI) is more viable than the manual classification. Various sorts of AI calculations have proposed in writing like K-Nearest Neighbors Principal Component Analysis, Gaussian Mixture Model, and Hidden Markov Model, etc. By utilizing the above methods, the audio classification can be done with no class related pre-information. However, they require huge training data with no real segregation results. To beat these insufficiencies, this paper proposed a general structure for audio classification. In this paper, another audio classification algorithm utilizing Support Vector Machine (SVM) in view of Whale Optimization Algorithm (WOA) is given where WOA-SVM utilizes the class mark of the info test as the real yield. WOA is utilized for conquering the inconvenience of SVM, for example, high computational multifaceted nature as a result of the explaining of enormous scale quadratic programming in parameter iterative learning methodology. The audio sign has shown up in huge volumes on account of its tendency. With the goal that we have utilized the MapReduce approach which is one of the sorts of big data investigation to play out the classification on the unstructured information. The proposed audio classification algorithm has contrasted with a few existing classification algorithms with demonstrating its productivity and the exactness.

A Simple Approach for Mapping Forest Cover from Time Series of Satellite Data

Forest cover mapping based on multi-temporal satellite observations usually uses dozens of features as inputs, which requires huge training data and leads to many ill effects. In this paper, a simple but efficient approach was proposed to map forest cover from time series of satellite observations without using classifiers and training data. This method focuses on the key step of forest mapping, i.e., separation of forests from herbaceous vegetation, considering that the non-vegetated area can be easily identified by the annual maximum vegetation index. We found that the greenness of forests is generally stable during the maturity period, but a similar greenness plateau does not exist for herbaceous vegetation. It means that the mean greenness during the vegetation maturity period of forests should be larger than that of herbaceous vegetation, while its standard deviation should be smaller. A combination of these two features could identify forests with several thresholds. The proposed approach was demonstrated for mapping the extents of different forest types with MODIS observations. The results show that the overall accuracy ranges 91.92–95.34% and the Kappa coefficient is 0.84–0.91 when compared with the reference datasets generated from fine-resolution imagery of Google Earth. The proposed approach can greatly simplify the procedures of forest cover mapping.

Reduced-DOF Three-Dimensional STAP via Subarray Synthesis for Nonsidelooking Planar Array Airborne Radar

Compared with conventional two-dimensional space-time adaptive processing (2D-STAP) methods, the elevation-azimuth-Doppler three-dimensional space-time adaptive processing (3D-STAP) method has the advantage of suppressing nonstationary clutter. Thereby, it is suitable for nonsidelooking airborne radar (non-SLAR) applications. However, its huge training data requirements and computational load are often beyond radar's ability in practical clutter environments. In this correspondence, we develop a simple but efficient reduced-degree-of-freedom (DOF) 3D-STAP method that significantly reduces the required training data and the computational complexity while maintaining the suboptimal clutter suppression performance. The proposed method transforms the planar array data into linear array data in azimuth and in elevation, respectively, thereby beamforming an equivalent cross-shape array prior to STAP. In consequence, only a few spatial DOFs, including azimuth and elevation dimension, are used for STAP directly resulting in the potential advantage for nonstationary clutter suppression and drastically reducing training data requirements and computational load. Furthermore, the clutter rank estimation rules of the planar array and the transformed cross-shape array are derived, and the required elevation DOFs of the proposed method are further discussed in detail. Simulations for clutter suppression of non-SLAR show that the proposed STAP method outperforms the state-of-the-art 3D-STAP method in terms of convergence and computational complexity.

Earlier detection of rumors in online social networks using certainty-factor-based convolutional neural networks

The rampant propagation of rumors in online social networks leads to potential damage to society. This phenomenon has attracted significant attention to the researches on faster rumor detection. Most recent works on rumor detection adapted the neural network-based deep learning approaches because of its high success rate. Though neural network-based models have high success rates, these models are less efficient in finding rumors at the earliest. Such models require huge training data for better and accurate results. Unfortunately, the data available in the early stages of rumor propagation are sparse in nature. This nature makes rumor detection using neural networks a complex solution, therefore rendering less efficiency for early stage rumors. To address this issue, we have proposed a certainty-factor-based convolutional neural network (CNN) approach to efficiently classify events as rumor or not by leveraging the inherent features of the set of information in spite of data sparsity. The certainty-factor-based activation function requires a minimum number of training data to obtain a generalization. In the proposed approach, two parallel CNNs are employed for the rumor event classification task that efficiently utilizes the inherent features of information, such as temporal, content, and propagation features. A decision tree combines the outputs of both CNNs and provides the classification output. This rumor event classification approach is then compared with recent and well-known state-of-the-art rumor classification approaches, and the results prove that the proposed approach detects rumor efficiently and rapidly with minimal inputs compared to other approaches.

State and Parameter Estimation for Spark Ignition Engine

In order to monitor the health of an internal combustion engine, there is a need to check the engine parameters from time to time. These parameters are further used in control and calibration. In this work, mean value engine dynamics of spark ignition model are taken and parameter estimation is done using Recursive Least Squares with forgetting factor. To get the information of unknown states, Unscented Kalman Filter is used for state estimation. Nonlinearity of the system is transferred by distributions. The parameters are estimated as time varying signals without having to define their dynamic behavior or as a function of other variables. This work explores a new combination of Unscented Kalman Filter and Recursive Least Squares with forgetting factor to estimate the physics based model parameters of spark ignition engine. This methodology outperforms many other methodologies by the fact that it does not require huge training data, parameter equations and restores the system nonlinearity. Simulation results are provided for Spark Ignition engine model with constant and time varying parameters. Comparison with Unscented Kalman Filter for the joint engine state and parameter estimation is done.