Research on classification method of high myopic maculopathy based on retinal fundus images and optimized ALFA-Mix active learning algorithm

AIM:To conduct a classification study of high myopic maculopathy(HMM)using limited datasets,including tessellated fundus,diffuse chorioretinal atrophy,patchy chorioretinal atrophy,and macular atrophy,and minimize annotation costs,and to optimize the ALFA-Mix active learning algorithm and apply it to HMM classification.METHODS:The optimized ALFA-Mix algorithm(ALFAMix+)was compared with five algorithms,including ALFA-Mix.Four models,including Res Net18,were established.Each algorithm was combined with four models for experiments on the HMM dataset.Each experiment consisted of 20 active learning rounds,with 100 images selected per round.The algorithm was evaluated by comparing the number of rounds in which ALFA-Mix+outperformed other algorithms.Finally,this study employed six models,including Efficient Former,to classify HMM.The best-performing model among these models was selected as the baseline model and combined with the ALFA-Mix+algorithm to achieve satisfactor y classification results with a small dataset.RESULTS:ALFA-Mix+outperforms other algorithms with an average superiority of 16.6,14.75,16.8,and 16.7 rounds in terms of accuracy,sensitivity,specificity,and Kappa value,respectively.This study conducted experiments on classifying HMM using several advanced deep learning models with a complete training set of 4252 images.The Efficient Former achieved the best results with an accuracy,sensitivity,specificity,and Kappa value of 0.8821,0.8334,0.9693,and 0.8339,respectively.Therefore,by combining ALFA-Mix+with Efficient Former,this study achieved results with an accuracy,sensitivity,specificity,and Kappa value of 0.8964,0.8643,0.9721,and 0.8537,respectively.CONCLUSION:The ALFA-Mix+algorithm reduces the required samples without compromising accuracy.Compared to other algorithms,ALFA-Mix+outperforms in more rounds of experiments.It effectively selects valuable samples compared to other algorithms.In HMM classification,combining ALFA-Mix+with Efficient Former enhances model performance,further demonstrating the effectiveness of ALFA-Mix+.

Dynamic Weapon Target Assignment Based on Intuitionistic Fuzzy Entropy of Discrete Particle Swarm

Aiming at the problems of convergence-slow and convergence-free of Discrete Particle Swarm Optimization Algorithm(DPSO) in solving large scale or complicated discrete problem, this article proposes Intuitionistic Fuzzy Entropy of Discrete Particle Swarm Optimization(IFDPSO) and makes it applied to Dynamic Weapon Target Assignment(WTA). First, the strategy of choosing intuitionistic fuzzy parameters of particle swarm is defined, making intuitionistic fuzzy entropy as a basic parameter for measure and velocity mutation. Second, through analyzing the defects of DPSO, an adjusting parameter for balancing two cognition, velocity mutation mechanism and position mutation strategy are designed, and then two sets of improved and derivative algorithms for IFDPSO are put forward, which ensures the IFDPSO possibly search as much as possible sub-optimal positions and its neighborhood and the algorithm ability of searching global optimal value in solving large scale 0-1 knapsack problem is intensified. Third, focusing on the problem of WTA, some parameters including dynamic parameter for shifting firepower and constraints are designed to solve the problems of weapon target assignment. In addition, WTA Optimization Model with time and resource constraints is finally set up, which also intensifies the algorithm ability of searching global and local best value in the solution of WTA problem. Finally, the superiority of IFDPSO is proved by several simulation experiments. Particularly, IFDPSO, IFDPSO1~IFDPSO3 are respectively effective in solving large scale, medium scale or strict constraint problems such as 0-1 knapsack problem and WTA problem.

An Investigation of Medical Students’ Cognition on the Doctor-Patient Relationship and Their Career Choice

Objective:To investigate medical students’cognition on the doctor-patient relationship and its influence on their career choice as well as to provide reference for medical education and communication education.Methods:A self-compiled questionnaire survey was carried out at a medical university in W city,and descriptive statistics as well as chi-square tests were conducted.Results:There were significant differences in medical students’cognition of doctor-patient relationship among different genders,registered residence,only-child status,whose parents are working in the medical field,and their own evaluation of the major(p<0.05);the different cognitions of doctor-patient relationship have a significant influence on medical students’career choice.Conclusion:Medical students’vocational cognition and educational guidance should be strengthened,hospital information should be open and transparent,media reporting of medical events should be standardized,and a doctor-patient conflict regulation mechanism should be established.

Spray Prediction Model for Aonla Rust Disease Using Machine Learning

Disease prediction in plants has acquired much attention in recent years.Meteorological factors such as:temperature,relative humidity,rainfall,sunshine play an important role in a plan’s growth only if they are present in adequate amounts as required by the plant.On the other hand,if the factors are inadequate,they may also support the growth of a disease in the plants.The current study focuses on the Rust disease in Aonla fruits and leaves by utilizing a real time dataset of weather parameters.Fifteen different models are tested for spray prediction on conducive days.Two resampling techniques,random over sampling(ROS)and synthetic minority oversampling technique(SMOTE)have been used to balance the dataset and five different classifiers:support vector machine(SVM),logistic regression(LR),k-nearest neighbor(kNN),decision tree(DT)and random forest(RF)have been used to classify a particular day based on weather conditions as conducive or non-conducive.The classifiers are then evaluated based on four performance metrics:accuracy,precision,recall and F1-score.The results indicate that for imbalanced dataset,kNN is appropriate with high precision and recall values.Considering both balanced and imbalanced dataset models,the proposed model SMOTE-RF performs best among all models with 94.6%accuracy and can be used in a real time application for spray prediction.Hence,timely fungicide spray prediction without over spraying will help in better productivity and will prevent the yield loss due to rust disease in Aonla crop.

A commentary of“Revealing the mechanism of photoperception regulating glucose metabolism”:Top 10 Scientific Advances of 2023,China

The widespread adoption of artificial light sources,particularly LED lights,catalyzed by industrial modernization,has revolutionized human living and working environments.This transformation,while enhancing life's convenience,harbors potential health implications.Public health studies have identified an association between nighttime light pollution and an increased risk of metabolic diseases such as diabetes and obesity.However,the biological mechanisms through which light regulates glucose metabolism remain unclear.

Effective and robust approach for fluorescence molecular tomography based on CoSaMP and SP_(3)model

Fluorescence molecular tomography(FMT)allows the detection and quantification of various biological processes in small animals in vrivo,which expands the horizons of pre clinical rescarch and drug development.Eficient three dimensional(3D)reconstruction algorithm is the key to accurate localization and quant ification of fAuorescent target in FMT.In this paper,3D recon-struction of FMT is regarded as a sparse signal recovery problem and the compressive sampling matching pursuit(CoSaMP)algorithm is adopted to obtain greedy recovery of fuorescent sig-nals.Moreover,to reduce the modeling error,the simplified spherical harmonics approximation to the radiative transfer equation(RTE),more specifically SP_(3),is utilized to describe light prop-agation in biological tissues.The performance of the proposed reconstruction method is thor-oughly evaluated by simulations on a 3D digital mouse model by comparing it with three representative greedy methods including orthogonal matching pursuit(OMP),stagewise OMP(StOMP),and regularized OMP(ROMP).The CoSaMP combined with SP_(3)shows an im-provement in reconstruction accuracy and exhibits distinct advantages over the comparative algorithms in multiple targets resolving.Stability analysis suggests that CoSaMP is robust to noise and performs stably with reduction of measurements.The feasibility and reoonstruction accuracy of the proposed method are further validated by phantom experimental data.

CFOA Based Low Pass and High Pass Ladder Filter—A New Configuration

A new technique using signal flow graph for conversion of ladder based filter into CFOA based filter has been proposed. The proposed technique converts the existing LC ladder based filter into CFOA in low pass and high pass configuration. The design of low pass filter and high pass filter has been realized using the proposed technique. The proposed configuration is implemented using CFOA as an active device and all the capacitors are grounded. Simulation has been carried out using simulation software I-cap. The simulation results have been demonstrated and discussed.

Reducing Carbon Emission through Container Shipment Consolidation and Optimization

Human’s impact on earth through global warming is more or less an accepted fact.Ocean freight is estimated to contribute 4-5%of global carbon emissions.Many manufacturing companies that transfer ship goods through full container loads found themselves under-utilizing the containers and resulting in higher carbon footprint per volume shipment.One of the reasons is the choice of non-ideal container sizes for their shipments.In this paper,we first provide an Integer Programming model to minimize the companies’shipping carbon footprints by selecting the ideal container sizes appropriate for their shipment volumes.Secondly,we proposed a strategy to minimize the carbon footprint by consolidating the shipments in the same country from multiple domestic locations at a port of loading by road freight,before the international sea shipment.A mixed-Integer Programming model has been developed to determine if one should ship each shipment separately or have shipments consolidated first before being shipped.Consolidation fills up the containers more efficiently that reduces the overall carbon footprint.Computational results using real-world data indicates a significant 13.4%reduction carbon emission when selecting the optimal combinations of different sizes of containers and an additional 12.1%reduction in carbon emission when shipment consolidation is applied.

从盲人摸象说起——谈谈对数字图书馆的认识

盲人摸象的故事之所以有教育意义,是因为我们都知道了大象是个什么样的动物,相比之下,盲人对大象的认识就显得很可笑.现实生活中,人类对某些新生事物在没有完全认知时,必须会经过像盲人摸象这一摸索阶段.在数字图书馆的教学研究领域,我们正处在盲人摸象的探索过程当中.

Pushing the Boundaries of Molecular Property Prediction for Drug Discovery with Multitask Learning BERT Enhanced by SMILES Enumeration

Accurate prediction of pharmacological properties of small molecules is becoming increasingly important in drug discovery.Traditional feature-engineering approaches heavily rely on handcrafted descriptors and/or fingerprints,which need extensive human expert knowledge.With the rapid progress of artificial intelligence technology,data-driven deep learning methods have shown unparalleled advantages over feature-engineering-based methods.However,existing deep learning methods usually suffer from the scarcity of labeled data and the inability to share information between different tasks when applied to predicting molecular properties,thus resulting in poor generalization capability.Here,we proposed a novel multitask learning BERT(Bidirectional Encoder Representations from Transformer)framework,named MTL-BERT,which leverages large-scale pre-training,multitask learning,and SMILES(simplified molecular input line entry specification)enumeration to alleviate the data scarcity problem.MTL-BERT first exploits a large amount of unlabeled data through self-supervised pretraining to mine the rich contextual information in SMILES strings and then fine-tunes the pretrained model for multiple downstream tasks simultaneously by leveraging their shared information.Meanwhile,SMILES enumeration is used as a data enhancement strategy during the pretraining,fine-tuning,and test phases to substantially increase data diversity and help to learn the key relevant patterns from complex SMILES strings.The experimental results showed that the pretrained MTL-BERT model with few additional fine-tuning can achieve much better performance than the state-of-the-art methods on most of the 60 practical molecular datasets.Additionally,the MTL-BERT model leverages attention mechanisms to focus on SMILES character features essential to target properties for model interpretability.

Near-field chiral excitation of universal spin-momentum locking transport of edge waves in microwave metamaterials

Controlling energy flow in waveguides has attractive potential in integrated devices from radio frequencies to optical bands.Due to the spin-orbit coupling,the mirror symmetry will be broken,and the handedness of the near-field source will determine the direction of energy transport.Compared with well-established theories about spin-momentum locking,experimental visualization of unidirectional coupling is usually challenging due to the lack of generic chiral sources and the strict environmental requirement.In this work,we design a broadband near-field chiral source in the microwave band and discuss experimental details to visualize spin-momentum locking in three different metamaterial waveguides,including spoof surface plasmon polaritons,line waves,and valley topological insulators.The similarity of these edge waves relies on the abrupt sign change of intrinsic characteristics of two media across the interface.In addition to the development of experimental technology,the advantages and research status of interface waveguides are summarized,and perspectives on future research are presented to explore an avenue for designing controllable spin-sorting devices in the microwave band.

Glide symmetry for mode control and significant suppression of coupling in dual-strip SSPP transmission lines

Glide symmetry,which is one kind of higher symmetry,is introduced in a special type of plasmonic metamaterial,the transmission lines(TLs)of spoof surface plasmon polaritons(SSPPs),in order to control the dispersion characteristics and modal fields of the SSPPs.We show that the glide-symmetric TL presents merged pass bands and mode degeneracy,which lead to broad working bandwidth and extremely low coupling between neighboring TLs.Dual-conductor SSPP TLs with and without glide symmetry are arranged in parallel as two channels with very deep subwavelength separation(e.g.,λ0∕100 at 5 GHz)for the application of integrated circuits and systems.Mutual coupling between the hybrid channels is analyzed using coupled mode theory and characterized in terms of scattering parameters and near-field distributions.We demonstrate theoretically and experimentally that the hybrid TL array obtains significantly more suppressed crosstalk than the uniform array of two nonglide symmetric TLs.Hence,it is concluded that the glide symmetry can be adopted to flexibly design the propagation of SSPPs and benefit the development of highly compact plasmonic circuits.

A robust adversarial attack against speech recognition with UAP

Speech recognition(SR)systems based on deep neural networks are increasingly widespread in smart devices.However,they are vulnerable to human-imperceptible adversarial attacks,which cause the SR to generate incorrect or targeted adversarial commands.Meanwhile,audio adversarial attacks are particularly susceptible to various factors,e.g.,ambient noise,after applying them to a real-world attack.To circumvent this issue,we develop a universal adversarial perturbation(UAP)generation method to construct robust real-world UAP by integrating ambient noise into the generation process.The proposed UAP can work well in the case of input-agnostic and independent sources.We validate the effectiveness of our method on two different SRs in different real-world scenarios and parameters,the results demonstrate that our method yields state-of-the-art performance,i.e.given any audio waveform,the word error rate can be up to 80%.Extensive experiments investigate the impact of different parameters(e.g,signal-to-noise ratio,distance,and attack angle)on the attack success rate.

Feature selection for face recognition:a memetic algorithmic approach

The eigenface method that uses principal component analysis(PCA) has been the standard and popular method used in face recognition.This paper presents a PCA-memetic algorithm(PCA-MA) approach for feature selection.PCA has been extended by MAs where the former was used for feature extraction/dimensionality reduction and the latter exploited for feature selection.Simulations were performed over ORL and YaleB face databases using Euclidean norm as the classifier.It was found that as far as the recognition rate is concerned,PCA-MA completely outperforms the eigenface method.We compared the performance of PCA extended with genetic algorithm(PCA-GA) with our proposed PCA-MA method.The results also clearly established the supremacy of the PCA-MA method over the PCA-GA method.We further extended linear discriminant analysis(LDA) and kernel principal component analysis(KPCA) approaches with the MA and observed significant improvement in recognition rate with fewer features.This paper also compares the performance of PCA-MA,LDA-MA and KPCA-MA approaches.

Light-controllable time-domain digital coding metasurfaces

Programmable metasurfaces enable real-time control of electromagnetic waves in a digital coding manner,which are suitable for implementing time-domain metasurfaces with strong harmonic manipulation capabilities.However,the time-domain metasurfaces are usually realized by adopting the wired electrical control method,which is effective and robust,but there are still some limitations.Here,we propose a light-controllable time-domain digital coding metasurface consisting of a full-polarization dynamic metasurface and a high-speed photoelectric detection circuit,from which the microwave reflection spectra are manipulated by time-varying light signals with periodic phase modulations.As demonstrated,the light-controllable timedomain digital coding metasurface is illuminated by the light signals with two designed time-coding sequences.The measured results show that the metasurface can well generate symmetrical harmonics and white-noise-like spectra,respectively,under such cases in the reflected wave.The proposed lightcontrollable time-varying metasurface offers a planar interface to tailor and link microwaves with lights in the time domain,which could promote the development of photoelectric hybrid metasurfaces and related multiphysics applications.

A Study on Iterative Learning Control for Nonlinear Impulsive Fractional Differential Systems

This paper explores P-type learning laws for impulsive nonlinear fractional differential systems.In order to prove the convergence of the system in P-type learning laws,we refer to the various properties of fractional systems and inequality.The sufficient conditions of iterative learning control is established.Finally,we give an example to illustrate theoretical results.