Low rank optimization for efficient deep learning:making a balance between compact architecture and fast training

Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices,and it is not environmental-friendly with much power cost.In this paper,we focus on low-rank optimization for efficient deep learning techniques.In the space domain,DNNs are compressed by low rank approximation of the network parameters,which directly reduces the storage requirement with a smaller number of network parameters.In the time domain,the network parameters can be trained in a few subspaces,which enables efficient training for fast convergence.The model compression in the spatial domain is summarized into three categories as pre-train,pre-set,and compression-aware methods,respectively.With a series of integrable techniques discussed,such as sparse pruning,quantization,and entropy coding,we can ensemble them in an integration framework with lower computational complexity and storage.In addition to summary of recent technical advances,we have two findings for motivating future works.One is that the effective rank,derived from the Shannon entropy of the normalized singular values,outperforms other conventional sparse measures such as the?_1 norm for network compression.The other is a spatial and temporal balance for tensorized neural networks.For accelerating the training of tensorized neural networks,it is crucial to leverage redundancy for both model compression and subspace training.

Transparent and Accountable Training Data Sharing in Decentralized Machine Learning Systems

In Decentralized Machine Learning(DML)systems,system participants contribute their resources to assist others in developing machine learning solutions.Identifying malicious contributions in DML systems is challenging,which has led to the exploration of blockchain technology.Blockchain leverages its transparency and immutability to record the provenance and reliability of training data.However,storing massive datasets or implementing model evaluation processes on smart contracts incurs high computational costs.Additionally,current research on preventing malicious contributions in DML systems primarily focuses on protecting models from being exploited by workers who contribute incorrect or misleading data.However,less attention has been paid to the scenario where malicious requesters intentionally manipulate test data during evaluation to gain an unfair advantage.This paper proposes a transparent and accountable training data sharing method that securely shares data among potentially malicious system participants.First,we introduce a blockchain-based DML system architecture that supports secure training data sharing through the IPFS network.Second,we design a blockchain smart contract to transparently split training datasets into training and test datasets,respectively,without involving system participants.Under the system,transparent and accountable training data sharing can be achieved with attribute-based proxy re-encryption.We demonstrate the security analysis for the system,and conduct experiments on the Ethereum and IPFS platforms to show the feasibility and practicality of the system.

A deep learning method based on prior knowledge with dual training for solving FPK equation

The evolution of the probability density function of a stochastic dynamical system over time can be described by a Fokker–Planck–Kolmogorov(FPK) equation, the solution of which determines the distribution of macroscopic variables in the stochastic dynamic system. Traditional methods for solving these equations often struggle with computational efficiency and scalability, particularly in high-dimensional contexts. To address these challenges, this paper proposes a novel deep learning method based on prior knowledge with dual training to solve the stationary FPK equations. Initially, the neural network is pre-trained through the prior knowledge obtained by Monte Carlo simulation(MCS). Subsequently, the second training phase incorporates the FPK differential operator into the loss function, while a supervisory term consisting of local maximum points is specifically included to mitigate the generation of zero solutions. This dual-training strategy not only expedites convergence but also enhances computational efficiency, making the method well-suited for high-dimensional systems. Numerical examples, including two different two-dimensional(2D), six-dimensional(6D), and eight-dimensional(8D) systems, are conducted to assess the efficacy of the proposed method. The results demonstrate robust performance in terms of both computational speed and accuracy for solving FPK equations in the first three systems. While the method is also applicable to high-dimensional systems, such as 8D, it should be noted that computational efficiency may be marginally compromised due to data volume constraints.

Personalized assessment and training of neurosurgical skills in virtual reality:An interpretable machine learning approach

Background Virtual reality technology has been widely used in surgical simulators,providing new opportunities for assessing and training surgical skills.Machine learning algorithms are commonly used to analyze and evaluate the performance of participants.However,their interpretability limits the personalization of the training for individual participants.Methods Seventy-nine participants were recruited and divided into three groups based on their skill level in intracranial tumor resection.Data on the use of surgical tools were collected using a surgical simulator.Feature selection was performed using the Minimum Redundancy Maximum Relevance and SVM-RFE algorithms to obtain the final metrics for training the machine learning model.Five machine learning algorithms were trained to predict the skill level,and the support vector machine performed the best,with an accuracy of 92.41%and Area Under Curve value of 0.98253.The machine learning model was interpreted using Shapley values to identify the important factors contributing to the skill level of each participant.Results This study demonstrates the effectiveness of machine learning in differentiating the evaluation and training of virtual reality neurosurgical performances.The use of Shapley values enables targeted training by identifying deficiencies in individual skills.Conclusions This study provides insights into the use of machine learning for personalized training in virtual reality neurosurgery.The interpretability of the machine learning models enables the development of individualized training programs.In addition,this study highlighted the potential of explanatory models in training external skills.

Design of a Student Recommendation Platform Based on Learning Behavior and Habit Training

This study innovatively built an intelligent analysis platform for learning behavior,which deeply integrated the cutting-edge technology of big data and Artificial Intelligence(AI),\mined and analyzed students’learning data,and realized the personalized customization of learning resources and the accurate matching of intelligent learning partners.With the help of advanced algorithms and multi-dimensional data fusion strategies,the platform not only promotes positive interaction and collaboration in the learning environment but also provides teachers with comprehensive and in-depth students’learning portraits,which provides solid support for the implementation of precision education and the personalized adjustment of teaching strategies.In this study,a recommender system based on user similarity evaluation and a collaborative filtering mechanism is carefully designed,and its technical architecture and implementation process are described in detail.

Boosting Adversarial Training with Learnable Distribution

In recent years,various adversarial defense methods have been proposed to improve the robustness of deep neural networks.Adversarial training is one of the most potent methods to defend against adversarial attacks.However,the difference in the feature space between natural and adversarial examples hinders the accuracy and robustness of the model in adversarial training.This paper proposes a learnable distribution adversarial training method,aiming to construct the same distribution for training data utilizing the Gaussian mixture model.The distribution centroid is built to classify samples and constrain the distribution of the sample features.The natural and adversarial examples are pushed to the same distribution centroid to improve the accuracy and robustness of the model.The proposed method generates adversarial examples to close the distribution gap between the natural and adversarial examples through an attack algorithm explicitly designed for adversarial training.This algorithm gradually increases the accuracy and robustness of the model by scaling perturbation.Finally,the proposed method outputs the predicted labels and the distance between the sample and the distribution centroid.The distribution characteristics of the samples can be utilized to detect adversarial cases that can potentially evade the model defense.The effectiveness of the proposed method is demonstrated through comprehensive experiments.

Safety-Constrained Multi-Agent Reinforcement Learning for Power Quality Control in Distributed Renewable Energy Networks

This paper examines the difficulties of managing distributed power systems,notably due to the increasing use of renewable energy sources,and focuses on voltage control challenges exacerbated by their variable nature in modern power grids.To tackle the unique challenges of voltage control in distributed renewable energy networks,researchers are increasingly turning towards multi-agent reinforcement learning(MARL).However,MARL raises safety concerns due to the unpredictability in agent actions during their exploration phase.This unpredictability can lead to unsafe control measures.To mitigate these safety concerns in MARL-based voltage control,our study introduces a novel approach:Safety-ConstrainedMulti-Agent Reinforcement Learning(SC-MARL).This approach incorporates a specialized safety constraint module specifically designed for voltage control within the MARL framework.This module ensures that the MARL agents carry out voltage control actions safely.The experiments demonstrate that,in the 33-buses,141-buses,and 322-buses power systems,employing SC-MARL for voltage control resulted in a reduction of the Voltage Out of Control Rate(%V.out)from0.43,0.24,and 2.95 to 0,0.01,and 0.03,respectively.Additionally,the Reactive Power Loss(Q loss)decreased from 0.095,0.547,and 0.017 to 0.062,0.452,and 0.016 in the corresponding systems.

Olive Leaf Disease Detection via Wavelet Transform and Feature Fusion of Pre-Trained Deep Learning Models

Olive trees are susceptible to a variety of diseases that can cause significant crop damage and economic losses.Early detection of these diseases is essential for effective management.We propose a novel transformed wavelet,feature-fused,pre-trained deep learning model for detecting olive leaf diseases.The proposed model combines wavelet transforms with pre-trained deep-learning models to extract discriminative features from olive leaf images.The model has four main phases:preprocessing using data augmentation,three-level wavelet transformation,learning using pre-trained deep learning models,and a fused deep learning model.In the preprocessing phase,the image dataset is augmented using techniques such as resizing,rescaling,flipping,rotation,zooming,and contrasting.In wavelet transformation,the augmented images are decomposed into three frequency levels.Three pre-trained deep learning models,EfficientNet-B7,DenseNet-201,and ResNet-152-V2,are used in the learning phase.The models were trained using the approximate images of the third-level sub-band of the wavelet transform.In the fused phase,the fused model consists of a merge layer,three dense layers,and two dropout layers.The proposed model was evaluated using a dataset of images of healthy and infected olive leaves.It achieved an accuracy of 99.72%in the diagnosis of olive leaf diseases,which exceeds the accuracy of other methods reported in the literature.This finding suggests that our proposed method is a promising tool for the early detection of olive leaf diseases.

Training Low-Achiever EFL Learners with Metacognitive Strategy Training for Effective Vocabulary Retention

There is ample research on the vocabulary learning strategies used by EFL and ESL learners.Moreover,many researchers have attempted to discern the effectiveness of teaching vocabulary learning strategies instruction.However,the impact of using metacognitive strategies along with vocabulary learning strategies on weak language learners has not been researched.The aim of this paper is to report on the effects of training low-achiever English as a foreign language learners with vocabulary learning strategies along with metacognitive strategies on vocabulary acquisition.Two B1 level groups of students were assigned as treatment and control groups at a university in Istanbul,Turkey.A Vocabulary Strategy Use Survey was given to identify the vocabulary strategy use of the subjects at the outset of the study.The treatment group was trained on vocabulary learning strategies combined with metacognitive strategy training.The control group studied the same words without any training.A vocabulary test was administered to both groups as a post-test and the results were compared.The findings of the Post-Test demonstrated that training weak language learners with vocabulary learning strategies along with metacognitive strategies has a positive impact on helping these learners increase their lexical knowledge,as the experimental group scored higher on the post-test compared to the control group.

Transfer Learning Approach to Classify the X-Ray Image that Corresponds to Corona Disease Using ResNet50 Pre-Trained by ChexNet

The COVID-19 pandemic has had a widespread negative impact globally. It shares symptoms with other respiratory illnesses such as pneumonia and influenza, making rapid and accurate diagnosis essential to treat individuals and halt further transmission. X-ray imaging of the lungs is one of the most reliable diagnostic tools. Utilizing deep learning, we can train models to recognize the signs of infection, thus aiding in the identification of COVID-19 cases. For our project, we developed a deep learning model utilizing the ResNet50 architecture, pre-trained with ImageNet and CheXNet datasets. We tackled the challenge of an imbalanced dataset, the CoronaHack Chest X-Ray dataset provided by Kaggle, through both binary and multi-class classification approaches. Additionally, we evaluated the performance impact of using Focal loss versus Cross-entropy loss in our model.

Implicit policy constraint for offline reinforcement learning

Offline reinforcement learning(RL)aims to learn policies entirely from passively collected datasets,making it a data‐driven decision method.One of the main challenges in offline RL is the distribution shift problem,which causes the algorithm to visit out‐of‐distribution(OOD)samples.The distribution shift can be mitigated by constraining the divergence between the target policy and the behaviour policy.However,this method can overly constrain the target policy and impair the algorithm's performance,as it does not directly distinguish between in‐distribution and OOD samples.In addition,it is difficult to learn and represent multi‐modal behaviour policy when the datasets are collected by several different behaviour policies.To overcome these drawbacks,the au-thors address the distribution shift problem by implicit policy constraints with energy‐based models(EBMs)rather than explicitly modelling the behaviour policy.The EBM is powerful for representing complex multi‐modal distributions as well as the ability to distinguish in‐distribution samples and OODs.Experimental results show that their method significantly outperforms the explicit policy constraint method and other base-lines.In addition,the learnt energy model can be used to indicate OOD visits and alert the possible failure.

Constrained Multi-Objective Optimization With Deep Reinforcement Learning Assisted Operator Selection

Solving constrained multi-objective optimization problems with evolutionary algorithms has attracted considerable attention.Various constrained multi-objective optimization evolutionary algorithms(CMOEAs)have been developed with the use of different algorithmic strategies,evolutionary operators,and constraint-handling techniques.The performance of CMOEAs may be heavily dependent on the operators used,however,it is usually difficult to select suitable operators for the problem at hand.Hence,improving operator selection is promising and necessary for CMOEAs.This work proposes an online operator selection framework assisted by Deep Reinforcement Learning.The dynamics of the population,including convergence,diversity,and feasibility,are regarded as the state;the candidate operators are considered as actions;and the improvement of the population state is treated as the reward.By using a Q-network to learn a policy to estimate the Q-values of all actions,the proposed approach can adaptively select an operator that maximizes the improvement of the population according to the current state and thereby improve the algorithmic performance.The framework is embedded into four popular CMOEAs and assessed on 42 benchmark problems.The experimental results reveal that the proposed Deep Reinforcement Learning-assisted operator selection significantly improves the performance of these CMOEAs and the resulting algorithm obtains better versatility compared to nine state-of-the-art CMOEAs.

Exploring the Application Effect of Flipped Classroom Combined with Problem-Based Learning Teaching Method in Clinical Skills Teaching of Standardized Training for Resident Doctors of Traditional Chinese Medicine

Objective: To explore the application effect of flipped classroom combined with problem-based learning teaching method in clinical skills teaching of standardized training for resident doctors of traditional Chinese Medicine. Methods: The study used the experimental control method. The study lasted from September to November 2022. The subjects of this study were 49 students of standardized training for resident doctors of traditional Chinese Medicine from grades 2020, 2021 and 2022 of Dazhou integrated TCM & Western Medicine Hospital. They were randomly divided into experiment group (25) and control group (24). The experiment group adopted flipped classroom combined with problem-based learning teaching method, and the control group adopted traditional teaching method. The teaching content was 4 basic clinical skill projects, including four diagnoses of traditional Chinese Medicine, cardiopulmonary resuscitation, dressing change procedure, acupuncture and massage. The evaluation method was carried out by comparing the students’ performance and a self-designed questionnaire was used to investigate the students’ evaluation of the teaching method. Results: The test scores of total scores in the experimental group (90.12 ± 5.89) were all higher than those in the control group (81.47 ± 7.96) (t = 4.53, P P Conclusions: The teaching process of the flipped classroom combined with problem-based learning teaching method is conducive to improving the efficiency of classroom teaching, cultivating students’ self-learning ability, and enhancing students’ willingness to learn.

Comparative Analysis of Machine Learning Models for PDF Malware Detection:Evaluating Different Training and Testing Criteria

The proliferation of maliciously coded documents as file transfers increase has led to a rise in sophisticated attacks.Portable Document Format(PDF)files have emerged as a major attack vector for malware due to their adaptability and wide usage.Detecting malware in PDF files is challenging due to its ability to include various harmful elements such as embedded scripts,exploits,and malicious URLs.This paper presents a comparative analysis of machine learning(ML)techniques,including Naive Bayes(NB),K-Nearest Neighbor(KNN),Average One Dependency Estimator(A1DE),RandomForest(RF),and SupportVectorMachine(SVM)forPDFmalware detection.The study utilizes a dataset obtained from the Canadian Institute for Cyber-security and employs different testing criteria,namely percentage splitting and 10-fold cross-validation.The performance of the techniques is evaluated using F1-score,precision,recall,and accuracy measures.The results indicate that KNNoutperforms other models,achieving an accuracy of 99.8599%using 10-fold cross-validation.The findings highlight the effectiveness of ML models in accurately detecting PDF malware and provide insights for developing robust systems to protect against malicious activities.

Deep learning-based battery state of charge estimation:Enhancing estimation performance with unlabelled training samples

The estimation of state of charge(SOC)using deep neural networks(DNN)generally requires a considerable number of labelled samples for training,which refer to the current and voltage pieces with knowing their corresponding SOCs.However,the collection of labelled samples is costly and time-consuming.In contrast,the unlabelled training samples,which consist of the current and voltage data with unknown SOCs,are easy to obtain.In view of this,this paper proposes an improved DNN for SOC estimation by effectively using both a pool of unlabelled samples and a limited number of labelled samples.Besides the traditional supervised network,the proposed method uses an input reconstruction network to reformulate the time dependency features of the voltage and current.In this way,the developed network can extract useful information from the unlabelled samples.The proposed method is validated under different drive cycles and temperature conditions.The results reveal that the SOC estimation accuracy of the DNN trained with both labelled and unlabelled samples outperforms that of only using a limited number of labelled samples.In addition,when the dataset with reduced number of labelled samples to some extent is used to test the developed network,it is found that the proposed method performs well and is robust in producing the model outputs with the required accuracy when the unlabelled samples are involved in the model training.Furthermore,the proposed method is evaluated with different recurrent neural networks(RNNs)applied to the input reconstruction module.The results indicate that the proposed method is feasible for various RNN algorithms,and it could be flexibly applied to other conditions as required.

A Novel Scheme for Separate Training of Deep Learning-Based CSI Feedback Autoencoders

In this paper, we introduce a novel scheme for the separate training of deep learning-based autoencoders used for Channel State Information (CSI) feedback. Our distinct training approach caters to multiple users and base stations, enabling independent and individualized local training. This ensures the more secure processing of data and algorithms, different from the commonly adopted joint training method. To maintain comparable performance with joint training, we present two distinct training methods: separate training decoder and separate training encoder. It’s noteworthy that conducting separate training for the encoder can pose additional challenges, due to its responsibility in acquiring a compressed representation of underlying data features. This complexity makes accommodating multiple pre-trained decoders for just one encoder a demanding task. To overcome this, we design an adaptation layer architecture that effectively minimizes performance losses. Moreover, the flexible training strategy empowers users and base stations to seamlessly incorporate distinct encoder and decoder structures into the system, significantly amplifying the system’s scalability. .

A Holistic Approach to Enhance FRCS General Surgery Examination Training Using Adult Learning Model: A Non-Profit Initiative

Objectives: To evaluate the impact of adult learning and simulation-based learning (SBL) on surgical trainees’ learning experiences and Fellowship of the Royal College of Surgeons (FRCS) Section 2 General Surgery examination pass rate. Methods: This was a cross-sequential study involving 148 surgical candidates (72 UK trainees, 75 non-UK trainees) who had attended our revision course (Phoenix FRCS Course) from June 2017 until 2023. Each course comprised a two-day weekend preparation with dedicated sections for clinical, viva, and academic reading, incorporating SBL as its key learning style. We maintained a prospective database of candidate and course details, examination results, and feedback since the course inception. Results: We found that 97% of candidates passed the FRCS examination after their first attempt. The course was attended once by 89% of candidates, and only 3 of the 148 candidates exhausted all four attempts at the examination. Candidate feedback for the course and its style of learning was positive, with simulation-based table viva sessions and virtual clinical sessions proving the most popular learning sessions (95% and 80% of candidates attending courses run in December 2017, April 2018, and May 2021 rated them “Excellent” respectively). Conclusions: The course is centered around shared adult learning and mindfulness tools to encourage candidates to learn from each other and develop confidence and mastery in all domains of surgical practice. These methods have been shown to be effective in achieving high success rates in the Intercollegiate and International FRCS examinations for UK and overseas surgeons.

Integration of Theory and Practice in Medical Morphology Curriculum in Postgraduate Training:A Flipped Classroom and Case-based Learning Exercise

Objective:The integration of training in theory and practice across the medical education spectrum is being encouraged to increase student understanding and skills in the sciences.This study aimed to determine the deciding factors that drive students'perceived advantages in class to improve precision education and the teaching model.Methods:A mixed strategy of an existing flipped classroom(FC)and a case-based learning(CBL)model was conducted in a medical morphology curriculum for 575 postgraduate students.The subjective learning evaluation of the individuals(learning time,engagement,study interest and concentration,and professional integration)was collected and analyzed after FC-CBL model learning.Results:The results from the general evaluation showed promising results of the medical morphology in the FC-CBL model.Students felt more engaged by instructors in person and benefited in terms of time-saving,flexible arrangements,and professional improvement.Our study contributed to the FC-CBL model in Research Design in postgraduate training in 4 categories:1)advancing a guideline of precision teaching according to individual characteristics;2)revealing whether a learning background is needed for a Research Design course to guide setting up a preliminary course;3)understanding the perceived advantages and their interfaces;and 4)barriers and/or improvement to implement the FC-CBL model in the Research Design class,such as a richer description of e-learning and hands-on practice.Conclusion:Undertaking a FC-CBL combined model could be a useful addition to pedagogy for medical morphology learning in postgraduate training.

M-learning结合CBL在消化科规培教学中的探讨及应用

目的探究移动学习平台(M-learning,ML)结合案例教学(Case-based Learning,CBL)在消化科住院医师规范化培训(简称规培)教学中的应用效果。方法选取2021年1月—2023年1月于上海市徐汇区中心医院消化科参加规培学习的80名医师作为研究对象,将其按照随机数表法分为研究组和对照组,每组40名。对照组给予传统讲授式教学法,研究组给予M-learning结合CBL教学法,对比两组医师的理论考试成绩、实践技能考试成绩和学习满意度。结果研究组的理论成绩和实践技能考试成绩均高于对照组,差异具有统计学意义(P均<0.05);研究组的学习满意度明显高于对照组,差异具有统计学意义(P<0.05)。结论将Mlearning结合CBL教学法应用于消化科规培教学中,不仅能够提升医师的理论考试成绩和实践技能考试成绩,还能够有效提高医师学习满意度。

基于改进Q-Learning的移动机器人路径规划算法

随着移动机器人在生产生活中的深入应用,其路径规划能力也需要向快速性和环境适应性兼备发展。为解决现有移动机器人使用强化学习方法进行路径规划时存在的探索前期容易陷入局部最优、反复搜索同一区域,探索后期收敛率低、收敛速度慢的问题,本研究提出一种改进的Q-Learning算法。该算法改进Q矩阵赋值方法,使迭代前期探索过程具有指向性,并降低碰撞的情况;改进Q矩阵迭代方法,使Q矩阵更新具有前瞻性,避免在一个小区域中反复探索;改进随机探索策略,在迭代前期全面利用环境信息,后期向目标点靠近。在不同栅格地图仿真验证结果表明,本文算法在Q-Learning算法的基础上,通过上述改进降低探索过程中的路径长度、减少抖动并提高收敛的速度,具有更高的计算效率。