Batch Input Research Articles

Convergence analysis of a novel high order networks model based on entropy error function

It is generally known that the error function is one of the key factors that determine the convergence, stability and generalization ability of neural networks. For most feedforward neural networks, the squared error function is usually chosen as the error function to train the network. However, networks based on the squared error function can lead to slow convergence and easily fall into local optimum in the actual training process. Recent studies have found that, compared to the squared error function, the gradient method based on the entropy error function measures the difference between the probability distribution of the model output and the probability distribution of the true labels during the iterative process, which can be more able to handle the uncertainty in the classification problem, less likely to fall into a local optimum and can learn to converge more rapidly. In this paper, we propose a batch gradient method for Sigma-Pi-Sigma neural networks based on the entropy error function and rigorously demonstrate the weak and strong convergence of the new algorithm in the batch input mode. Finally, the theoretical results and effectiveness of the algorithm are verified by simulation.

Optimal Controller Identification for multivariable non-minimum phase systems

This work deals with data-driven control for non-minimum phase (NMP) systems, where the goal is to design a controller for a plant whose model is unknown by using a batch of input–output data collected from it. The approach is based on the Model Reference paradigm, where a transfer function matrix – the reference model – is used to specify the desired closed-loop performance. The NMP systems issue in Model Reference approaches is a well-known problem in control literature and it is no different in data-driven methods. This work explains how to adapt the formulation of the Optimal Controller Identification (OCI) method to cope with this class of systems. Considering a convenient parametrization of the reference model and a flexible performance criterion, it is possible to identify the NMP transmission zeros of the plant along with the optimal controller parameters, as it will be shown. Both diagonal and block-triangular reference model structures are treated in detail. Simulation examples show the effectiveness of the proposed approach.

Transfer-Aware Graph U-Net with Cross-Level Interactions for PolSAR Image Semantic Segmentation

Although graph convolutional networks have found application in polarimetric synthetic aperture radar (PolSAR) image classification tasks, the available approaches cannot operate on multiple graphs, which hinders their potential to generalize effective feature representations across different datasets. To overcome this limitation and achieve robust PolSAR image classification, this paper proposes a novel end-to-end cross-level interaction graph U-Net (CLIGUNet), where weighted max-relative spatial convolution is proposed to enable simultaneous learning of latent features from batch input. Moreover, it integrates weighted adjacency matrices, derived from the symmetric revised Wishart distance, to encode polarimetric similarity into weighted max-relative spatial graph convolution. Employing end-to-end trainable residual transformers with multi-head attention, our proposed cross-level interactions enable the decoder to fuse multi-scale graph feature representations, enhancing effective features from various scales through a deep supervision strategy. Additionally, multi-scale dynamic graphs are introduced to expand the receptive field, enabling trainable adjacency matrices with refined connectivity relationships and edge weights within each resolution. Experiments undertaken on real PolSAR datasets show the superiority of our CLIGUNet with respect to state-of-the-art networks in classification accuracy and robustness in handling unknown imagery with similar land covers.

Analysis of tensile damage of titanium alloy in seawater environment based on deep learning

With the continuous expansion of the global ocean economy, the demand for marine engineering is increasing. However, the presence of chemicals such as chloride ions in the marine environment poses a significant corrosion risk to metal materials, thereby compromising the safety and reliability of marine engineering projects. This study focused on investigating the corrosion treatment and mechanical performance of titanium alloy materials. Simulated seawater corrosion experiments and integrated acoustic emission systems were utilized to gather mechanical performance and acoustic emission data from 48 samples after corrosion treatment. The study applied consistency treatment to the acoustic emission-remaining useful life (AE-RUL) data and established a correlation between acoustic emission signals and the remaining life of the titanium alloy during the yield stage. Principal component analysis was employed to reduce the initial 11-dimensional data to 5 dimensions, improving the quality of the dataset. The Long Short Term Memory (LSTM) network was enhanced using interpolation optimization methods and multi-time-series batch input techniques. The predictions and reliability analysis for the tensile yield life of titanium alloy achieved an accuracy rate of 94.102 %, which was further improved to 95.782 % with the use of spline interpolation methods. Reliability analysis demonstrated a good fit with a normal distribution through parameter estimation and hypothesis testing within a 95 % confidence interval.

Winds of Change: Enhancing Wind Power Generation Forecasting with LSTM Models and Advanced Techniques in a Single Turbine Wind Farm

This article presents a comprehensive analysis of wind power generation using a Long Short-Term Memory (LSTM) model in the context of a wind farm with a single turbine. The novelty of the research lies in the integration of a technical model, predictive model, heat map analysis, and LSTM architecture to enhance forecasting accuracy. The performance of the forecasted model is evaluated using standard metrics, including Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). In the estimation phase, wind power generation is predicted based on current weather conditions and historical trends. Leveraging multivariate time series forecasting with LSTM through the Kera library, different look-back values are explored to optimise prediction accuracy. In particular, the study achieves favourable performance metrics, including MSE of 27.781, RMSE of 5.271, MAE of 3.281, and variance of 0.886 with a 70% training and 30% test split. Furthermore, a 60% training and 40% test split yield improved accuracy with MSE of 28.791, RMSE of 5.366, MAE of 3.376, and variance of 0.888. In the prediction phase, the LSTM model is employed to forecast power generation without relying on future weather information. Through various experiments, the optimal look-back period and number of neurones for the LSTM model are determined. Notably, the achieved mean absolute percent errors (MAPE) of 11.433% and 11.158% for 24 and 48 hours of data respectively showcase the impact of varying neuron counts on prediction accuracy. To further enhance predictive capabilities, optimisation techniques are implemented, involving adjustments to the LSTM model, including increased input batch size, batch normalisation, and dense layers. This optimisation leads to a substantial decrease in MAPE from 92.53% to 87.14% in a monthly prediction experiment. Furthermore, forward predictions are made for multiple days in the future, where LSTM successfully captures patterns and fluctuations, resulting in MAPEs of 70.74% for 24 hours, 39.26% for 2 days, and 51.48% for 1 week. Additionally, a comparison of two LSTM-based models, Autoencoder LSTM and FFT-Encoder-Decoder-LSTM, demonstrates the superior performance of the latter in terms of RMSE. This study emphasizes the potential of LSTM models in wind power generation prediction while introducing novel insights into optimizing model performance for accurate forecasting. The integration of technical and predictive models, along with heat map analysis and LSTM architecture, sets the groundwork for advancing wind power prediction methodologies.

Arrhythmia classification based on multi-feature multi-path parallel deep convolutional neural networks and improved focal loss.

Early diagnosis of abnormal electrocardiogram (ECG) signals can provide useful information for the prevention and detection of arrhythmia diseases. Due to the similarities in Normal beat (N) and Supraventricular Premature Beat (S) categories and imbalance of ECG categories, arrhythmia classification cannot achieve satisfactory classification results under the inter-patient assessment paradigm. In this paper, a multi-path parallel deep convolutional neural network was proposed for arrhythmia classification. Furthermore, a global average RR interval was introduced to address the issue of similarities between N vs. S categories, and a weighted loss function was developed to solve the imbalance problem using the dynamically adjusted weights based on the proportion of each class in the input batch. The MIT-BIH arrhythmia dataset was used to validate the classification performances of the proposed method. Experimental results under the intra-patient evaluation paradigm and inter-patient evaluation paradigm showed that the proposed method could achieve better classification results than other methods. Among them, the accuracy, average sensitivity, average precision, and average specificity under the intra-patient paradigm were 98.73%, 94.89%, 89.38%, and 98.24%, respectively. The accuracy, average sensitivity, average precision, and average specificity under the inter-patient paradigm were 91.22%, 89.91%, 68.23%, and 95.23%, respectively.

Machine learning model and input batch management tool for the composition of new recycled polypropylene plastic material with reduced variability in target properties

Reducing the variability of recycled plastic materials is a prerequisite to encourage the safe substitution of virgin materials in new components manufacturing, hence promoting the circular economy (CE) to achieve the recyclability objectives set by the European Commission. To increase the use of recycled materials in some industrial sectors, such as in automotive, it is necessary to control and reduce their batch-to-batch variability to ensure that the required properties lay in reasonable ranges. In this work, through the analysis of the accumulated historical data from a recycling company, and with the help of machine learning (ML) techniques, a model is developed that predicts key material properties (melt flow index, ash content, Izod impact strength and percentage of shrinkage) in a mixture of recycled polypropylene (PP) batches from different sources. The model is validated correlating predictions with experimental measurements over new batch mixtures compounded on purpose. A batch management algorithm considers the entry flow of material batches received in the recycling company and proposes mixtures for the production of new recycled batches with reduced variability and controllable target properties. Furthermore, it proposes solutions for batch management to maximize the production output while maintaining the properties of the recycled material within target variability limits.

Exploring the Potential of ChatGPT-4 in Predicting Refractive Surgery Categorizations: Comparative Study.

Refractive surgery research aims to optimally precategorize patients by their suitability for various types of surgery. Recent advances have led to the development of artificial intelligence-powered algorithms, including machine learning approaches, to assess risks and enhance workflow. Large language models (LLMs) like ChatGPT-4 (OpenAI LP) have emerged as potential general artificial intelligence tools that can assist across various disciplines, possibly including refractive surgery decision-making. However, their actual capabilities in precategorizing refractive surgery patients based on real-world parameters remain unexplored. This exploratory study aimed to validate ChatGPT-4's capabilities in precategorizing refractive surgery patients based on commonly used clinical parameters. The goal was to assess whether ChatGPT-4's performance when categorizing batch inputs is comparable to those made by a refractive surgeon. A simple binary set of categories (patient suitable for laser refractive surgery or not) as well as a more detailed set were compared. Data from 100 consecutive patients from a refractive clinic were anonymized and analyzed. Parameters included age, sex, manifest refraction, visual acuity, and various corneal measurements and indices from Scheimpflug imaging. This study compared ChatGPT-4's performance with a clinician's categorizations using Cohen κ coefficient, a chi-square test, a confusion matrix, accuracy, precision, recall, F1-score, and receiver operating characteristic area under the curve. A statistically significant noncoincidental accordance was found between ChatGPT-4 and the clinician's categorizations with a Cohen κ coefficient of 0.399 for 6 categories (95% CI 0.256-0.537) and 0.610 for binary categorization (95% CI 0.372-0.792). The model showed temporal instability and response variability, however. The chi-square test on 6 categories indicated an association between the 2 raters' distributions (χ²5=94.7, P<.001). Here, the accuracy was 0.68, precision 0.75, recall 0.68, and F1-score 0.70. For 2 categories, the accuracy was 0.88, precision 0.88, recall 0.88, F1-score 0.88, and area under the curve 0.79. This study revealed that ChatGPT-4 exhibits potential as a precategorization tool in refractive surgery, showing promising agreement with clinician categorizations. However, its main limitations include, among others, dependency on solely one human rater, small sample size, the instability and variability of ChatGPT's (OpenAI LP) output between iterations and nontransparency of the underlying models. The results encourage further exploration into the application of LLMs like ChatGPT-4 in health care, particularly in decision-making processes that require understanding vast clinical data. Future research should focus on defining the model's accuracy with prompt and vignette standardization, detecting confounding factors, and comparing to other versions of ChatGPT-4 and other LLMs to pave the way for larger-scale validation and real-world implementation.

SCORE: approximating curvature information under self-concordant regularization

Optimization problems that include regularization functions in their objectives are regularly solved in many applications. When one seeks second-order methods for such problems, it may be desirable to exploit specific properties of some of these regularization functions when accounting for curvature information in the solution steps to speed up convergence. In this paper, we propose the SCORE (self-concordant regularization) framework for unconstrained minimization problems which incorporates second-order information in the Newton-decrement framework for convex optimization. We propose the generalized Gauss–Newton with Self-Concordant Regularization (GGN-SCORE) algorithm that updates the minimization variables each time it receives a new input batch. The proposed algorithm exploits the structure of the second-order information in the Hessian matrix, thereby reducing computational overhead. GGN-SCORE demonstrates how to speed up convergence while also improving model generalization for problems that involve regularized minimization under the proposed SCORE framework. Numerical experiments show the efficiency of our method and its fast convergence, which compare favorably against baseline first-order and quasi-Newton methods. Additional experiments involving non-convex (overparameterized) neural network training problems show that the proposed method is promising for non-convex optimization.

Semi-supervised deep neural networks for left atrial segmentation from cardiac MRI

Abstract Funding Acknowledgements Type of funding sources: None. Introduction Left atrium size and function is an important but less studied prognostic marker for many cardiovascular diseases [1]. Cardiac Magnetic Resonance Imaging (MRI) can capture atrial structure with a high temporal and spatial resolution. However, diagnoses and accurate measurement of volumes from MRI often rely on time-consuming manual processing, or needs correction of software segmentation error by clinicians. Robust, automated methods are highly desirable to assist morphological analysis of dynamic changes in the heart. Ultimately this facilitates the wider application of cardiac MRI in early identification and prevention of potential cardiovascular disease. Purpose To introduce a semi-supervised teacher-student deep neural networks (requiring less labelled data) for accurate segmentation of left atrium from cardiac MRI images, and compare their performance with conventional supervised learning. Methods The basic structure for a mean teacher deep neural network is shown in Fig.1. Network training starts in a supervised manner. At each MRI batch input, both the student and the teacher model evaluate the images with applied noise. The student network is updated by gradient descent and segmentation loss is obtained by comparing its prediction with the label. The exponential moving average (EMA) of the student weight is then migrated to update the teacher network. Consistency loss between the student’s and teacher’s prediction (filtered by uncertainty map in Uncertainty-Aware Mean Teacher, UA-MT [2]) is obtained to jointly update the student. Unlabeled images continually improve the network without using supervised loss. We compared three semi-supervised teacher-student networks UA-MT, Hierarchical Consistency Regularized Mean Teacher (HCR-MT) and Semi-supervised Contrastive Consistency model (SCC), with the supervised network TransUNnet [3], using benchmark cardiac MRI dataset from MICCAI 2018 Atrial Segmentation Challenge [4]. Out of 20 MRI scans (1760 images), 16 scans were used for training, and 4 scans for testing. For semi-supervised learning, 75% of the training set was used as labelled data and 25% was assumed to be unlabeled. In the supervised learning, all labels were enabled. Results Example segmentation results from three semi-supervised models are shown in Fig 2. For comparison, the fully supervised TransUNnet produced an accuracy of 76%. Benefiting from effective teacher-student learning, the semi-supervised models achieved a superior accuracy of 89%, while also requiring less labelled data. Conclusion The proposed teacher-student deep neural network facilitates effective semi-supervised learning, and outperforms supervised TransUNet in segmenting the left atrium in cardiac MRI. Our semi-supervised approach uses less training data, reducing tedious and time-consuming manual labelling tasks, ultimately unlocking the ability to effectively utilise the full power of deep learning in medical applications.

Fold3D: Rethinking and Parallelizing Computational and Communicational Tasks in the Training of Large DNN Models

Training a large DNN (e.g., GPT3) efficiently on commodity clouds is challenging even with the latest 3D parallel training systems (e.g., Megatron v3.0). In particular, along the pipeline parallelism dimension, computational tasks that produce a whole DNN's gradients with multiple input batches should be concurrently activated; along the data parallelism dimension, a set of heavy-weight communications (for aggregating the accumulated outputs of computational tasks) is <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">inevitably serialized</i> after the pipelined tasks, undermining the training performance (e.g., in Megatron, data parallelism caused all GPUs idle for over 44% of the training time) over commodity cloud networks. To deserialize these communicational and computational tasks, we propose the AIAO scheduling (for 3D parallelism) which slices a DNN into multiple segments, so that the computational tasks processing the same DNN segment can be scheduled together, and the communicational tasks that synchronize this segment can be launched and overlapped (deserialized) with other segments’ computational tasks. We realized this idea in our <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Fold3D</small> training system. Extensive evaluation shows <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Fold3D</small> eliminated most of the all-GPU 44% idle time in Megatron (caused by data parallelism), leading to 25.2%–42.1% training throughput improvement compared to four notable baselines over various settings; <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Fold3D</small> 's high performance scaled to many GPUs.

Dislocation Substructures Evolution and an Informer Constitutive Model for a Ti-55511 Alloy in Two-Stages High-Temperature Forming with Variant Strain Rates in β Region.

The high-temperature compression characteristics of a Ti-55511 alloy are explored through adopting two-stage high-temperature compressed experiments with step-like strain rates. The evolving features of dislocation substructures over hot, compressed parameters are revealed by transmission electron microscopy (TEM). The experiment results suggest that the dislocations annihilation through the rearrangement/interaction of dislocations is aggravated with the increase in forming temperature. Notwithstanding, the generation/interlacing of dislocations exhibit an enhanced trend with the increase in strain in the first stage of forming, or in strain rates at first/second stages of a high-temperature compressed process. According to the testing data, an Informer deep learning model is proposed for reconstructing the stress-strain behavior of the researched Ti-55511 alloy. The input series of the established Informer deep learning model are compression parameters (compressed temperature, strain, as well as strain rate), and the output series are true stresses. The optimal input batch size and sequence length are 64 and 2, respectively. Eventually, the predicted results of the proposed Informer deep learning model are more accordant with the tested true stresses compared to those of the previously established physical mechanism model, demonstrating that the Informer deep learning model enjoys an outstanding forecasted capability for precisely reconstructing the high-temperature compressed features of the Ti-55511 alloy.

Quality-Aware Joint Caching, Computing and Communication Optimization for Video Delivery in Vehicular Networks

As being clear that the joint optimization of caching, computing, and communication (3 C) can significantly improve vehicular network performance, this paper investigates the quality-aware 3 C optimization for video delivery in vehicular networks. By incorporating the quality-awareness into 3 C models, we formulate both the quality-aware joint 3 C optimization problem and the quality-aware joint computing and communication (2 C) optimization problem with predetermined caching. By observing that the proposed 2 C and 3 C problems can be interpreted as submodular maximization problems with constraints, we propose a solution methodology using the submodularity and knapsack interpretation. Based on the proposed methodology, we then propose efficient solution approaches assuming batch input and streaming input, respectively. Complexity of the proposed solution approaches is analyzed. In addition, extensive simulations are conducted to evaluate the proposed solution approaches. Results show that our proposed approaches can outperform the reference schemes and improve the video delivery of vehicular networks.

Automatic detection of pesticide residues on the surface of lettuce leaves using images of feature wavelengths spectrum.

The inappropriate application of pesticides to vegetable crops often results in environmental pollution, which seriously impacts the environment and human health. Given that current methods of pesticide residue detection are associated with issues such as low accuracy, high equipment cost, and complex flow, this study puts forward a new method for detecting pesticide residues on lettuce leaves. To establish this method, spectral analysis was used to determine the characteristic wavelength of pesticide residues (709 nm), machine vision equipment was improved, and a bandpass filter and light source of characteristic wavelength were installed to acquire leaf image information. Next, image preprocessing and feature information extraction were automatically implemented through programming. Several links were established for the training model so that the required feature information could be automatically extracted after the batch input of images. The pesticide residue detected using the chemical method was taken as the output and modeled, together with the input image information, using the convolutional neural network (CNN) algorithm. Furthermore, a prediction program was rewritten to generalize the input images during the prediction process and directly obtain the output pesticide residue. The experimental results revealed that when the detection device and method designed in this study were used to detect pesticide residues on lettuce leaves in a key state laboratory, the coefficient of determination of the equation reached 0.883, and the root mean square error (RMSE) was 0.134 mg/L, indicating high accuracy and that the proposed method integrated the advantages of spectrum detection and deep learning. According to comparison testing, the proposed method can meet Chinese national standards in terms of accuracy. Moreover, the improved machine vision equipment was less expensive, thus providing powerful support for the application and popularization of the proposed method.

Towards energy efficient scheduling in the olive oil extraction industry: Comparative assessment of energy consumption in two management models

• Energy Use Efficiency and Specific Energy in olive oil mill was evaluated. • Batch and continuous extraction processes were compared. • The Overall Equipment Effectiveness in batch and continuous olive mill was evaluated. • The energy consumed in the batch system is double that of the continuous one. • Batch system performance is nearly half of continuous one. Although olive oil is one of the most important food products throughout the Mediterranean region, the main consideration in configuring and operating extraction plants is still the quantity of crop that will be brought by the individual growers, who are often of small and medium size. On this basis, the common configuration is the batch processing line (Ba-PL) for small and medium capacities, with malaxers arranged in parallel, or in larger-capacity industrial mills, the arrangement of malaxers in series for simulation of continuous processing (Co-PL). A review of literature to date reveals that there have been no assessments of the energy use and scheduling strategies for these two processes, therefore the current study undertakes such assessment and comparison of the Ba-PL and Co-PL configurations, within a single mill facility. The results show that energy output for Ba-PL was 105,570.00 MJ day −1 based on inputs of 3212.76 MJ day −1 , while outputs from Co-PL reached 422,280.00 MJ day −1 from inputs of 6740.38 MJ day −1 . Given that the yield of oil is almost the same in the two processes, the Net Energy (NE) for Ba-Pl then results as 10,2357.24 MJ day −1 versus 41,5539.62 MJ day −1 for the Co-PL configuration, and moreover the later process is seen to achieve almost double the Energy Use Efficiency, at 62.65 versus 32.86 in Ba-PL. Finally, the Overall Equipment Effectiveness (OEE) analysis of performance for the two processes shows values of 93.1 % for the Co-PL configuration but only 51.2 % for Ba-PL, due mainly to dead times in throughput on the Ba-PL machinery, resulting in a 44.7 % decrease in the values for the Performance parameter. The conclusion is that to meet the needs of small producers, there must be new management approaches and improvement in the energy use for batch processing. Some strategies could be: – measure the oil content and moisture in input batches using rapid, non-destructive methods, so that homogeneous batches can be pooled before processing; – reduce dead time by developing automated systems for pumping the olive paste from the malaxers to the separation decanter; – even in the case that they are non-homogenous, consider pooling undersized batches so as to avoid mass flows far below the malaxer capacities. Throughout the Mediterranean, olive oil production using batch processing remains very substantial, therefore further studies are required for the identification and evaluation of solutions for optimising this process.

Ex 2 : Monte Carlo Tree Search‐based test inputs prioritization for fuzzing deep neural networks

Fuzzing is considered to be an essential approach to guarantee the reliability of deep neural networks (DNNs) based systems. The DNN fuzzing leverages various inputs prioritization methods to guide the testing process. The current research mainly focus on constructing testing metrics that symbolize the logical representation of the DNN to guide the generation of test cases, which neglects the potential performance brought by implementing heuristic algorithm. Moreover, the straightforward implementation of queue structure can not represent the metamorphic relationships between generated inputs in DNN fuzzing. Therefore, developing the appropriate heuristic algorithm-based inputs prioritization method is critical to improve the performance of DNN fuzzers. In this paper, we propose a Monte Carlo Tree Search (MCTS) based inputs prioritization method called E x 2 $E{x}^{2}$ (Exploration and Exploitation) that formulates DNN testing exploration as the sequential decision process. The technique introduces an innovative tree-structure design that schedules inputs from the statistical perspective. Different from traditional DNN testing, the batch pool is maintained in the form of nodes in MCTS. The links between nodes precisely represent the metamorphic relationship between input batches, which indicates the potential value for in-depth search. Furthermore, a novel simulation mechanism is implemented to adapt MCTS in DNN testing, which attain better coverage feedback. The effectiveness of our method is comprehensively investigated on six popular deep learning models from LeNet and VGG families. The comparison experiments are conducted between DeepHunter, TensorFuzz, and DeepSmartFuzzer to demonstrate efficacy on various testing metrics. The experimental results show that the E x 2 $E{x}^{2}$ significantly enhance the coverage gain of DNN fuzzing up to 30% against the best performance in comparison groups.

Static and Dynamic Deflection Characteristics of Cracked Porous FG Panels

The deflection responses of the cracked porous graded structure under the variable loading conditions (static and dynamic) are investigated numerically with the help of a commercial tool (ABAQUS). The model includes the grading in two different directions (along the thickness and length by setting the relevant volume fraction exponents [Formula: see text] and [Formula: see text], i.e., of the functionally graded material (FGM) for the analysis purpose. Additionally, different material gradings, i.e., power-law (P-FGM), sigmoid (S-FGM), and exponential (E-FGM), are adopted to achieve the distribution patterns within the FG panel. Further, the porosity and the damage (crack) have been introduced in the graded structural panel. The python scripting language is used to model the FGM plate via the batch input technique, i.e., ABAQUS kernel, instead of the standard GUI interface. The accuracy has been confirmed by validating the results with the published available results, and the applicability is stated through a series of numerical examples by changing input parameters.

Interpretable Optimization Training Strategy-Based DCNN and Its Application on CT Image Recognition

The theoretical basis of the discrete random sample batch classification is not clear and the sample division is not scientific during the process of Deep Convolutional Neural Network (DCNN) model training. Aiming at the problems above, starting from the DCNN detection recognition mechanism, the theory of random discrete samples is given and proved, and a scientific quantitative batch of sample input method is proposed. Combined with image preprocessing, based on the strategy of random dispersion of samples, and scientifically quantified sample input batches, the DCNN model is trained with limited label samples, and then the CT image recognition of pulmonary nodules is carried out. Experimental results based on the LIDC-ID-RI public dataset show that the sensitivity, specificity, and accuracy of the proposed method have reached 96.40%, 95.60%, and 96.00%, respectively. Compared with the multiscale convolutional neural network method and the multiscale multimode image fusion method, the recognition accuracy of the proposed method is improved by 1.6 and 3.49 percentage points, respectively.

CAMDNN: Content-Aware Mapping of a Network of Deep Neural Networks on Edge MPSoCs

Machine Learning (ML) workloads are increasingly deployed at the edge. Enabling efficient inference execution while considering model and system heterogeneity remains challenging, especially for ML tasks built with a network of DNNs. The challenge is to maximize the utilization of all available resources on the multiprocessor system on a chip (MPSoC) at the same time. This becomes even more complicated because the optimal mapping for the network of DNNs can vary with input batch sizes and scene complexity. In this paper, a holistic hierarchical scheduling framework is presented to optimize the execution time for a network of DNN models on an edge MPSoC at runtime, considering varying input characteristics. The framework consists of a local and a global scheduler. The local scheduler maps individual DNNs in the inference pipeline to the best-performing hardware unit while the global scheduler customizes an Integer Linear Programming (ILP) solution to instantiate DNN remapping. To minimize scheduler runtime overhead, an imitation learning (IL) based scheduler is used that approximates the ILP solutions. The proposed scheduling framework (CAMDNN) was implemented on a Qualcomm Robotic RB5 platform. CAMDNN resulted in lower execution time of up to 32% than HEFT, and by factors of 6.67X, 5.6X and 2.17X than the CPU-only, GPU-only and Central Queue schedulers.

Category Trees - Classifiers that Branch on Category

This paper presents a batch classifier that splits a dataset into tree branches depending on the category type. It has been improved from the earlier version and fixed a mistake in the earlier paper. Two important changes have been made. The first is to represent each category with a separate classifier. Each classifier then classifies its own subset of data rows, using batch input values to create the centroid and also represent the category itself. If the classifier contains data from more than one category however, it needs to create new classifiers for the incorrect data. The second change therefore is to allow the classifier to branch to new layers when there is a split in the data, and create new classifiers there for the data rows that are incorrectly classified. Each layer can therefore branch like a tree - not for distinguishing features, but for distinguishing categories. The paper then suggests a further innovation, which is to represent some data columns with fixed value ranges, or bands. When considering features, it is shown that some of the data can be classified directly through fixed value ranges, while the rest must be classified using a classifier technique and the idea allows the paper to discuss a biological analogy with neurons and neuron links. Tests show that the method can successfully classify a diverse set of benchmark datasets to better than the state-of-the-art.