DeepLearning4J深度学习框架及应用

DeepLearning4J是Skymind开源并托管在Eclipse基金会的深度分布式学习框架,是支持构建、训练和部署神经网络的内置框架学习神经网络工具包,目前,DeepLearning4J(以下简称DL4J)深度学习已经成为了Java商用平台应用热点。由一个经典的机器学习算法例子入手,简析机器学习算法在DL4J框架的实现,并通过在Windows系统下搭建训练测试环境,定型并保存手写体字符识别的DL4J模型,实现手写字符的识别,从而实现DeepLearning4J分布式深度学习库的学习与应用。

基于扩散模型的拓扑优化研究

拓扑优化是工业设计领域中常见的数学方法,旨在给定的物理领域内,满足各种约束条件、负载和其他边界条件等前提下,生成最佳的拓扑结构。传统的拓扑优化大多都依赖有限元方法(FEM),然而有限元方法的迭代计算很大程度上增加了拓扑优化的时间成本和算力成本。如今,机器学习和深度学习在图像生成领域内的快速发展为拓扑优化的发展带来了机遇。扩散模型是一种无监督图像生成模型,因生成效果优秀、细节完美等特点等得到广泛使用。本文将在扩散模型的基础上提出新的网络结构,让其适应拓扑优化生成特性,根据特定信息生成与之对应的最优拓扑优化结果。

A Novel Deep Learning Representation for Industrial Control System Data

Feature extraction plays an important role in constructing artificial intel-ligence(AI)models of industrial control systems(ICSs).Three challenges in this field are learning effective representation from high-dimensional features,data heterogeneity,and data noise due to the diversity of data dimensions,formats and noise of sensors,controllers and actuators.Hence,a novel unsupervised learn-ing autoencoder model is proposed for ICS data in this paper.Although traditional methods only capture the linear correlations of ICS features,our deep industrial representation learning model(DIRL)based on a convolutional neural network can mine high-order features,thus solving the problem of high-dimensional and heterogeneous ICS data.In addition,an unsupervised denoising autoencoder is introduced for noisy ICS data in DIRL.Training the denoising autoencoder allows the model to better mitigate the sensor noise problem.In this way,the represen-tative features learned by DIRL could help to evaluate the safety state of ICSs more effectively.We tested our model with absolute and relative accuracy experi-ments on two large-scale ICS datasets.Compared with other popular methods,DIRL showed advantages in four common indicators of AI algorithms:accuracy,precision,recall,and F1-score.This study contributes to the effective analysis of large-scale ICS data,which promotes the stable operation of ICSs.

A System of Image Recognition-Based Railway Foreign Object Intrusion Monitoring Design

The monitoring system designed in this paper is on account of YOLOv5(You Only Look Once)to monitor foreign objects on railway tracks and can broadcast the monitoring information to the locomotive in real time.First,the general structure of the system is determined through demand analysis and feasibility analysis,the foreign object intrusion recognition algorithm is designed,and the data set required for foreign object intrusion recognition is made.Secondly,according to the functional demands,the system selects a suitable neural web,and the programming is reasonable.At last,the system is simulated to validate its functionality(identification and classification of track intrusion and determination of a safe operating zone).

Traffic Signal Timing via Deep Reinforcement Learning

In this paper, we propose a set of algorithms to design signal timing plans via deep reinforcement learning. The core idea of this approach is to set up a deep neural network(DNN) to learn the Q-function of reinforcement learning from the sampled traffic state/control inputs and the corresponding traffic system performance output. Based on the obtained DNN,we can find the appropriate signal timing policies by implicitly modeling the control actions and the change of system states.We explain the possible benefits and implementation tricks of this new approach. The relationships between this new approach and some existing approaches are also carefully discussed.

Prediction Model Using Reinforcement Deep Learning Technique for Osteoarthritis Disease Diagnosis

Osteoarthritis is the most common class of arthritis that involves tears down the soft cartilage between the joints of the knee.The regeneration of this cartilage tissue is not possible,and thus physicians typically suggest therapeutic measures to prevent further deterioration over time.Normally,bringing about joint replacement is a remedial course of action.Expose itself in joint pain recog-nized with a normal X-ray.Deep learning plays a vital role in predicting the early stages of osteoarthritis by using the MRI pictures of muscles of the knee muscle.It can be used to accurately measure the shape and texture of biological structures can be measured consistently from X-ray images.Moreover,deep learning-based computation can be used to design framework to predict whether a given patient will develop osteoarthritis.Such a framework can identify clear biochemical changes in the focal point of ligaments of the knees of patients who have exhibit pre-indications in standard imaging.This study proposes framework to identify cases of osteoarthritis by using deep learning and reinforcement learning.It can be used as a clinical mechanism to predict the occurrence of osteoarthritis so that patients can benefit from early intervention.

Understand Students Feedback Using Bi-Integrated CRF Model Based Target Extraction

Educational institutions showing interest to find the opinion of the students about their course and the instructors to enhance the teaching-learning process.For this,most research uses sentiment analysis to track students’behavior.Traditional sentence-level sentiment analysis focuses on the whole sentence sentiment.Previous studies show that the sentiments alone are not enough to observe the feeling of the students because different words express different sentiments in a sentence.There is a need to extract the targets in a given sentence which helps to find the sentiment towards those targets.Target extraction is the subtask of targeted sentiment analysis.In this paper,we proposed the innovative model to find the targets of the given sentence using Bi-Integrated Conditional Random Fields(CRF).A Parallel fusion neural network model is designed to perform this task.We evaluate the model using the Michigan dataset and we build a dataset for target extraction from student reviews.The experimental results show that our proposed fusion model achieves better results compared to baseline models.

DeepSI:A Sensitive-Driven Testing Samples Generation Method of Whitebox CNN Model for Edge Computing

In recent years,Deep Learning(DL)technique has been widely used in Internet of Things(IoT)and Industrial Internet of Things(IIoT)for edge computing,and achieved good performances.But more and more studies have shown the vulnerability of neural networks.So,it is important to test the robustness and vulnerability of neural networks.More specifically,inspired by layer-wise relevance propagation and neural network verification,we propose a novel measurement of sensitive neurons and important neurons,and propose a novel neuron coverage criterion for robustness testing.Based on the novel criterion,we design a novel testing sample generation method,named DeepSI,which involves definitions of sensitive neurons and important neurons.Furthermore,we construct sensitive-decision paths of the neural network through selecting sensitive neurons and important neurons.Finally,we verify our idea by setting up several experiments,then results show our proposed method achieves superior performances.

Deep learning models for automatic identification of plant-parasitic nematode

Plant-parasitic nematodes cause various diseases that can be fatal to the infected plants.It causes losses to the agricultural industry,such as crop failure and poor crop quality.Developing an accurate nematode classification system is vital for pest identification and control.Deep learning classification techniques can help speed up Nematode identification as it can perform tasks directly from images.In the present study,four state-of-the-art deep learning models(ResNet101v2,CoAtNet-0,Effi-cientNetV2B0,and EfficientNetV2M)were evaluated in plantparasitic nematode classification from microscopic image.The models were trained using a combination of three different optimizers(Adam,SGD,dan RMSProp)and several data augmentation with image transformations,such as image flip,blurring,noise addition,brightness,and contrast adjustment.The performance of the trained models was varied.Regarding test accuracy,EfficientNetV2B0 and EfficientNetV2M using RMSProp and brightness augmentation give the best result of 97.94%However,the overall performance of EfficientNetV2M was superior,with 98.66%mean class accuracy,97.99%F1 score,98.26%average precision,and 97.94%average recall.

Progress in research on ultrasound radiomics for predicting the prognosis of breast cancer

Breast cancer is the most common malignant tumor and the leading cause of cancer-related deaths in women worldwide.Effective means of predicting the prognosis of breast cancer are very helpful in guiding treatment and improving patients'survival.Features extracted by radiomics reflect the genetic and molecular characteristics of a tumor and are related to its biological behavior and the patient's prognosis.Thus,radiomics provides a new approach to noninvasive assessment of breast cancer prognosis.Ultrasound is one of the commonest clinical means of examining breast cancer.In recent years,some results of research into ultrasound radiomics for diagnosing breast cancer,predicting lymph node status,treatment response,recurrence and survival times,and other aspects,have been published.In this article,we review the current research status and technical challenges of ultrasound radiomics for predicting breast cancer prognosis.We aim to provide a reference for radiomics researchers,promote the development of ultrasound radiomics,and advance its clinical application.

Deep Learning-Assisted Visualized Fluorometric Sensor Array for Biogenic Amines Detection

Biogenic amines(BAs)are important biomarkers for monitoring food quality and assisting in the diagnosis of disease.Facial,portable,accurate and high-throughput BAs detection is still challenging by the specific sensor compounds development or the complicated instrument operation.Deep learning(DL)algorithms are blooming for their superiority on the nonlinear and multidimensional data analysis,which endow the great advantage for the artificial intelligence assisted large sample analysis of the environmental or daily health monitoring.In this work,we developed a deep learning-assisted visualized fluorometric array-based sensing method.Two commercial fluorescent dyes were selected and combined into sensor arrays.Variation in the alkalinity of BAs causes significant and distinct fluorescence changes of the dyes.In conjunction with pattern recognition by the pretrained CNN models,the sensor array clearly differentiates seven BAs with 99.29%prediction accuracy and allows rapid single and multi-component quantification with a volume fraction range from 200 cm^(3)/m^(3)to 2500 cm^(3)/m^(3).This method also provides a new way for meat freshness monitoring.We envision that this novel analytical method for BAs can be used as an alternative and promising tool for the detection of a wider variety of analytes.

Deeper Attention-Based Network for Structured Data

Deep learning methods are applied into structured data and in typical methods,low-order features are discarded after combining with high-order featuresfor prediction tasks.However,in structured data,ignorance of low-order features may cause the low prediction rate.To address this issue,in this paper,deeper attention-based network(DAN)is proposed.With DAN method,to keep both low-and high-order features,attention average pooling layer was utilized to aggregate features of each order.Furthermore,by shortcut connections from each layer to attention average pooling layer,DAN can be built extremely deep to obtain enough capacity.Experimental results show DAN has good performance and works effectively.

基于深度学习的动漫人脸识别综述

近年来动漫产业的飞速发展、元宇宙的迅速火热以及NFT数字藏品的备受瞩目,让大家看到动漫产业的潜质与版权意识的觉醒。可对于人脸识别的研究大部分处于现实人脸识别,在动漫人脸识别领域的研究较少,但随上述产业的火热与深度学习技术的发展,现如今基于深度学习的动漫人脸识别也引起研究者们广泛地关注。