Robust Algorithm Research Articles

Benchmarking clustering, alignment, and integration methods for spatial transcriptomics

BackgroundSpatial transcriptomics (ST) is advancing our understanding of complex tissues and organisms. However, building a robust clustering algorithm to define spatially coherent regions in a single tissue slice and aligning or integrating multiple tissue slices originating from diverse sources for essential downstream analyses remains challenging. Numerous clustering, alignment, and integration methods have been specifically designed for ST data by leveraging its spatial information. The absence of comprehensive benchmark studies complicates the selection of methods and future method development.ResultsIn this study, we systematically benchmark a variety of state-of-the-art algorithms with a wide range of real and simulated datasets of varying sizes, technologies, species, and complexity. We analyze the strengths and weaknesses of each method using diverse quantitative and qualitative metrics and analyses, including eight metrics for spatial clustering accuracy and contiguity, uniform manifold approximation and projection visualization, layer-wise and spot-to-spot alignment accuracy, and 3D reconstruction, which are designed to assess method performance as well as data quality. The code used for evaluation is available on our GitHub. Additionally, we provide online notebook tutorials and documentation to facilitate the reproduction of all benchmarking results and to support the study of new methods and new datasets.ConclusionsOur analyses lead to comprehensive recommendations that cover multiple aspects, helping users to select optimal tools for their specific needs and guide future method development.

High-performance deep spiking neural networks with 0.3 spikes per neuron

Communication by rare, binary spikes is a key factor for the energy efficiency of biological brains. However, it is harder to train biologically-inspired spiking neural networks than artificial neural networks. This is puzzling given that theoretical results provide exact mapping algorithms from artificial to spiking neural networks with time-to-first-spike coding. In this paper we analyze in theory and simulation the learning dynamics of time-to-first-spike-networks and identify a specific instance of the vanishing-or-exploding gradient problem. While two choices of spiking neural network mappings solve this problem at initialization, only the one with a constant slope of the neuron membrane potential at threshold guarantees the equivalence of the training trajectory between spiking and artificial neural networks with rectified linear units. For specific image classification architectures comprising feed-forward dense or convolutional layers, we demonstrate that deep spiking neural network models can be effectively trained from scratch on MNIST and Fashion-MNIST datasets, or fine-tuned on large-scale datasets, such as CIFAR10, CIFAR100 and PLACES365, to achieve the exact same performance as that of artificial neural networks, surpassing previous spiking neural networks. Our approach accomplishes high-performance classification with less than 0.3 spikes per neuron, lending itself for an energy-efficient implementation. We also show that fine-tuning spiking neural networks with our robust gradient descent algorithm enables their optimization for hardware implementations with low latency and resilience to noise and quantization.

Theory and application of robust linear phase-shift algorithm for phase-shift deflectometry method

Based on the polynomial theory, the error propagation characteristics of the widely used N-step discrete Fourier transform (N-DFT) phase-shift algorithm were analyzed via theoretical analysis, under the effect of Gamma distortion and phase detuning. The results showed that the N-DFT algorithm could not simultaneously suppress both types of error. A robust linear phase-shift (RLPS) algorithm was designed, the performance of the RLPS and 8-DFT algorithms in terms of spectral response, detuning robustness, and GS/N\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${G}_{S/N}$$\\end{document} was briefly analysis by Manuel Servin method. The Simulation analysis and comparison of the results show that the RLPS algorithm could suppress both types of error simultaneously, which exhibited better stability and accuracy than N-DFT and exponential algorithms, particularly in gradient measurement stability, peak-to-valley (PV) and root-mean-square (RMS) error suppression. Moreover, a physical experiment apparatus was built to test unidirectionally inclined plane mirror and concave mirror using the RLPS, N-DFT, and exponential algorithms. The results showed that the RLPS algorithm could significantly improve the measurement stability and accuracy in the presence of detuning and without screen Gamma calibration.

Precision Cancer Medicine 2.0-Oncology in the postgenomic era.

Genomic medicine has transformed the lives of patients with cancer byenabling individualised and evidence-based clinical decision-making. Despite this progress, the implementation of precision cancer medicine is limited by its dependence on isolated biomarkers. The development of bulk and single-cell multiomic technologies has revealed the enormous complexity of the cancer ecosystem. Beyond the cancer cell, the tumour microenvironment, macroenvironment and host factors, including the microbiome, profoundly influence the cancer phenotype, and accounting for these enhances the resolution of precision medicine. The advent of robust multiomic profiling and interpretable machine learning algorithms mark the dawn of a new postgenomic era of personalised cancer medicine. In Precision Cancer Medicine 2.0, high-resolution personalised clinical decision-making is informed by the comprehensive multiomic profiling of tumour and host, integrated using artificial intelligence.

Investigating the Effect of ChatGPT-like New Generation AI Technology on User Entrepreneurial Activities

ChatGPT, characterized by its reliance on big data, robust algorithms, and significant computational power, has become a benchmark AI application product, signifying a new breakthrough in AI technology. The emergence of applications based on ChatGPT-like next-generation AI technology has triggered a series of interconnected transformations in human society's ways of thinking, production, living, and governance. However, the academic community has yet to conduct research specifically on innovation and entrepreneurship. Against this backdrop, this study explores the effect of the novel features of ChatGPT-like next-generation AI technology on user entrepreneurs, driving factors, and the entrepreneurial process. The findings reveal the following: (1) User entrepreneurs collect extensive user data through ChatGPT-like AI technology and intelligently analyze it to achieve optimal entrepreneurial judgments and decisions. (2) User entrepreneurs utilize ChatGPT-like AI technology to understand the latent needs of users and to acquire user demand information, such as product shortcomings and appeals. (3) ChatGPT-like AI technology enhances the entrepreneurial intention of user entrepreneurs, stimulates their creative thinking, and expands and deepens their social networks, thereby strengthening their identification with entrepreneurial opportunities. (4) ChatGPT-like AI technology drives and empowers the three-stage evolution of user entrepreneurship: idea generation, prototype development, and commercialization of innovative products. This study not only provides new insights and theoretical foundations for user entrepreneurship research to better explore and leverage the application of ChatGPT-like AI technology in the entrepreneurial process but also offers significant practical implications for encouraging users to actively engage in innovation and entrepreneurship activities, supporting the achievement of sustainable digital entrepreneurship goals.

Masketeer: An Ensemble-Based Pseudonymization Tool with Entity Recognition for German Unstructured Medical Free Text

Background: The recent rise of large language models has triggered renewed interest in medical free text data, which holds critical information about patients and diseases. However, medical free text is also highly sensitive. Therefore, de-identification is typically required but is complicated since medical free text is mostly unstructured. With the Masketeer algorithm, we present an effective tool to de-identify German medical text. Methods: We used an ensemble of different masking classes to remove references to identifiable data from over 35,000 clinical notes in accordance with the HIPAA Safe Harbor Guidelines. To retain additional context for readers, we implemented an entity recognition scheme and corpus-wide pseudonymization. Results: The algorithm performed with a sensitivity of 0.943 and specificity of 0.933. Further performance analyses showed linear runtime complexity (O(n)) with both increasing text length and corpus size. Conclusions: In the future, large language models will likely be able to de-identify medical free text more effectively and thoroughly than handcrafted rules. However, such gold-standard de-identification tools based on large language models are yet to emerge. In the current absence of such, we hope to provide best practices for a robust rule-based algorithm designed with expert domain knowledge.

Algorithm for Key Transparency with Transparent Logs

Background Cryptography plays a crucial role in securing digital communications and data storage. This study evaluates the Transparent Key Management Algorithm utilizing Merkle trees, focusing on its performance and security effectiveness in cryptographic key handling. Methods The research employs simulated experiments to systematically measure and analyze key operational metrics such as insertion and verification times. Synthetic datasets are used to mimic diverse operational conditions, ensuring rigorous evaluation under varying workloads and security threats. Implementation is carried out using R programming, integrating cryptographic functions and Merkle tree structures for integrity verification. Results Performance analysis reveals efficient insertion and verification times under normal conditions, essential for operational workflows. Security evaluations demonstrate the algorithm's robustness against tampering, with approximately 95% of keys verified successfully and effective detection of unauthorized modifications. Simulated attack scenarios underscore its resilience in mitigating security threats. Conclusions The Transparent Key Management Algorithm, enhanced by Merkle trees and cryptographic hashing techniques, proves effective in ensuring data integrity, security, and operational efficiency. Recommendations include continuous monitoring and adaptive algorithms to bolster resilience against evolving cybersecurity challenges, promoting trust and reliability in cryptographic operations.

Machine learning classification of permeable conducting spheres in air and seawater using electromagnetic pulses

Abstract This paper presents machine learning classification on simulated data of permeable conducting spheres in air and seawater irradiated by low frequency electromagnetic pulses. Classification accuracy greater than 90% was achieved. The simulated data were generated using an analytical model of a magnetic dipole in air and seawater placed 1.5–3.5 m above the center of the sphere in 50 cm increments. The spheres had radii of 40 cm and 50 cm and were of permeable materials, such as steel, and non-permeable materials, such as aluminum. A series RL circuit was analytically modeled as the transmitter coil, and an RLC circuit as the receiver coil. Additive white Gaussian noise was added to the simulated data to test the robustness of the machine learning algorithms to noise. Multiple machine learning algorithms were used for classification including a perceptron and multiclass logistic regression, which are linear models, and a neural network, 1D convolutional neural network (CNN), and 2D CNN, which are nonlinear models. Feature maps are plotted for the CNNs and provide explainability of the salient parts of the time signature and spectrogram data used for classification. The pulses investigated, which expand the literature, include a two-sided decaying exponential, Heaviside step-off, triangular, Gaussian, rectangular, modulated Gaussian, raised cosine, and rectangular down-chirp. Propagation effects, including dispersion and frequency dependent attenuation, are encapsulated by the analytical model, which was verified using finite element modeling. The results in this paper show that machine learning methods are a viable alternative to inversion of electromagnetic induction (EMI) data for metallic sphere classification, with the advantage of real-time classification without the use of a physics-based model. The nonlinear machine learning algorithms used in this work were able to accurately classify metallic spheres in seawater even with significant pulse distortion caused by dispersion and frequency dependent attenuation. This paper presents the first effort towards the use of machine learning to classify metallic objects in seawater based on EMI sensing.

Synthesis of eco-friendly polyaniline-zeolite nanocomposite for pollutant remediation: Empowered by robust machine learning algorithm

The present research assesses the synthesis of an eco-friendly polyaniline and zeolite 4A (4 Armstrong) nanocomposite for enhanced adsorption of dye and heavy metal. Zeolite has been synthesized utilizing natural Kutch-sourced Kaolin clay. The materials synthesized were characterized via various instrumental techniques which confirms the formation of the desired materials. The composite exhibits superior performance in removing hazardous pollutants, demonstrating the removal of 98.5 % chromium in 12 min and 99.8% Congo red in 50 min, individually. Furthermore, for simultaneous removal, the composite continued displaying remarkable performance having a removal rate of 99% and 98.5 % for Congo red and chromium, respectively within 50 min, only. The nanocomposite follows the Freundlich isotherm model for chromium removal indicating physisorption and the Langmuir isotherm model for Congo removal signifying chemisorption. Artificial intelligence and machine learning (ML) techniques have been employed on the data set for the simulation of the adsorption percentage of the synthesized adsorbent. Various uncharted data visualizing techniques, including scatter plots, correlation coefficients, and mutual information were applied to explain the correlation between experimental variables. Conventional regression techniques showed poor performance with a high mean squared error (MSE) of 23 and an R2 value of 0.89. However, Bayesian optimization assisted ML model significantly improved the results. The Gaussian process with a polynomial function was the best performer, achieving an R2 value of 0.99 and mean squared error of 0.008. This research highlights the novel synthesis and effectiveness of the adsorbent materials, demonstrating the potential of machine learning in environmental remediation.

Monitoring Welding Torch Position and Posture Using Reversed Electrode Images – Part II, Experimental and Analysis for the REI-TPA Model

This paper is a sequel to the previous paper. In the previous research (Ref. 1), a REI-TPA model was established to quantitively relate reversed electrode images (REIs) to welding torch position and attitude (TPA). In this research, the REI-TPA model was validated with bead-on-plate welding experiments on S304 stainless steel plates. The contours of the REI and electrode as well as the weld pool geometry were extracted from the image with a developed, robust algorithm, and the arc length was calculated with welding voltage. The offset distance and deflection angle of the welding torch relative to the correct position and attitude were calculated by inputting the extracted parameters into the REI-TPA model. The computational result was compared to the experimental data. The result showed that the model is correct and the monitoring of the welding torch with the REI-TPA model is available. The REITPA model can be applied to real-time control of TPA, which is a supplement to the application of the weld passive visual image and an extension of multiinformation acquisition and processing methods in the welding process.

Automatic driving image matching via Random Sample Consensus (RANSAC) and Spectral Clustering (SC) with monocular camera.

In today's big data era, with the development of the Internet of Things (IoT) technology and the trend of autonomous driving prevailing, visual information has shown a blowout increase, but most image matching algorithms have problems such as low accuracy and low inlier rates, resulting in insufficient information. In order to solve the problem of low image matching accuracy and low inlier rate in the field of autonomous driving, this research innovatively applies spectral clustering (SC) in the field of data analysis to image matching in the field of autonomous driving, and a new image matching algorithm "SC-RANSAC" based on SC and Random Sample Consensus (RANSAC) is proposed. The datasets in this research are collected based on the monocular cameras of autonomous driving cars. We use RANSAC to obtain the initial inlier set and the SC algorithm to filter RANSAC's outliers and then use the filtered inliers as the final inlier set. In order to verify the effectiveness of the algorithm, it shows the matching effect from three angles: camera translation, rotation, and rotation and translation. SC-RANSAC is also compared with RANSAC, graph-cut RANSAC, and marginalizing sample consensus by using two different types of datasets. Finally, we select three representative pictures to test the robustness of the SC-RANSAC algorithm. The experimental results show that SC-RANSAC can effectively and reliably eliminate mismatches in the initial matching results; has a high inlier rate, real-time performance, and robustness; and can be effectively applied in the environment of autonomous driving.

Hop-to-Hug algorithm: Novel strategy to stable cutting-plane algorithm based on convexification of yield functions

Numerical challenges, incorporating non-uniqueness, non-convexity, undefined gradients, and high curvature, of the positive level sets of yield function are encountered in stress integration when utilizing the return-mapping algorithm family. These phenomena are illustrated by an assessment of four typical yield functions: modified spatially mobilized plane criterion, Lade criterion, Bigoni-Piccolroaz criterion, and micromechanics-based upscaled Drucker-Prager criterion. One remedy to these issues, named the "Hop-to-Hug" (H2H) algorithm, is proposed via a convexification enhancement upon the classical cutting-plane algorithm (CPA). The improved robustness of the H2H algorithm is demonstrated through a series of integration tests in one single material point. Furthermore, a constitutive model is implemented with the H2H algorithm into the Abaqus/Standard finite-element platform. Element-level and structure-level analyses are carried out to validate the effectiveness of the H2H algorithm in convergence. All validation analyses manifest that the proposed H2H algorithm can offer enhanced stability over the classical CPA method while maintaining the ease of implementation, in which evaluations of the second-order derivatives of yield function and plastic potential function are circumvented.

Selective multiple kernel fuzzy clustering with locality preserved ensemble

Multiple kernel fuzzy clustering (MKFC) has demonstrated promising performance in capturing the non-linear relationships within data. However, its effectiveness relies heavily on the appropriate selection of the fuzzification coefficient and kernel functions. To address this challenge, this paper proposes a novel clustering ensemble approach for improving the robustness and accuracy of the MKFC algorithm. The proposed method employs a multi-objective evolutionary optimization approach to heuristically select the optimal fuzzification and kernel coefficients. By employing the selected coefficients, the MKFC algorithm generates a diverse set of accurate candidate clustering results. Subsequently, a locality preserved ensemble mechanism is introduced to derive the final partition matrix, ensuring that all candidate clusterings within the Pareto non-dominated set contribute to the final consensus matrix. Moreover, this mechanism incorporates the knowledge about the locality of the dataset via a graph regularization term, thereby further enhancing the clustering performance. Comprehensive experiments conducted on widely adopted benchmark datasets demonstrate the superiority of the proposed method over the state-of-the-art approaches.

Molecular separation-assisted label-free SERS combined with machine learning for nasopharyngeal cancer screening and radiotherapy resistance prediction

Nasopharyngeal cancer (NPC) is a malignant tumor with high prevalence in Southeast Asia and highly invasive and metastatic characteristics. Radiotherapy is the primary strategy for NPC treatment, however there is still lack of effect method for predicting the radioresistance that is the main reason for treatment failure. Herein, the molecular profiles of patient plasma from NPC with radiotherapy sensitivity and resistance groups as well as healthy group, respectively, were explored by label-free surface enhanced Raman spectroscopy (SERS) based on surface plasmon resonance for the first time. Especially, the components with different molecular weight sizes were analyzed via the separation process, helping to avoid the possible missing of diagnostic information due to the competitive adsorption. Following that, robust machine learning algorithm based on principal component analysis and linear discriminant analysis (PCA-LDA) was employed to extract the feature of blood-SERS data and establish an effective predictive model with the accuracy of 96.7% for identifying the radiotherapy resistance subjects from sensitivity ones, and 100% for identifying the NPC subjects from healthy ones. This work demonstrates the potential of molecular separation-assisted label-free SERS combined with machine learning for NPC screening and treatment strategy guidance in clinical scenario.

Promote computer vision applications in pig farming scenarios: high-quality dataset, fundamental models, and comparable performance1

Computer vision is widely recognized as an influential technology in the field of precision management of animals. Emerging studies have demonstrated the potential to improve pig health and welfare through animal surveillance systems and computer vision (CV) algorithms. However, the lack of benchmark datasets and robust fundamental algorithms restrict CV applications for the commercial use. This study aims to bridge the gap between technology development and commercial applications in pig farming scenarios by introducing a general-purpose dataset (PigLife), comparing benchmark performances of foundational CV algorithms and model development workflows. The PigLife dataset contains video clips and images (38 short video clips, 2K image frames, 22K pig instances) across most pig production phases in a typical commercial pig farm: Breeding and Gestation, Farrow to Wean, Weaning & Nursery, and Growth to Finish. Three detection algorithms (Faster R-CNN, RetinaNet, TridentNet) and three segmentation algorithms (Mask R-CNN, MViTv2, Point-Rend) were trained on the PigLife dataset from scratch. Fine-tuning of pre-trained models (YOLO8-m, Faster-RCNN-r50) and no-training from zero-shot models (CLIP-SAM, Grouddino-HQSAM) were also evaluated to suggest faster CV development workflows for commercial applications in pig farming. This study emphasizes the necessity of a benchmark dataset for evaluating the robustness of algorithms and identifying the remaining difficulties and challenges across various algorithms. Furthermore, developing CV models from pre-trained algorithms or zero-shot models showed better performance and a faster process, which could reduce barriers when developing high-performance CV products in pig production industry.

Islanded microgrid: hybrid energy resilience optimization

To maintain a dependable and sustainable power supply in a microgrid system, it is crucial to combine renewable energy sources with hybrid energy backup. To achieve maximum output power from solar source, a high gain DC-DC boost converter is managed using the dual Kalman filter based perturb and observe approach. A sliding mode current controller at a three phase inverter with an LC filter is intended to follow an undisturbed reference voltage. To effectively manage the power flow and optimize the utilization of available resources, a robust power management algorithm is required. The novel power management algorithm for a solo operated renewable distribution generation unit with hybrid energy backup in a microgrid is introduced. The algorithm aims to dynamically allocate power among various sources, storage systems, and loads, considering their characteristics and the overall system constraints. The algorithm utilizes sliding mode control techniques to regulate the current flow from the renewable generator and effectively manage the power allocation among different energy sources, storage systems, and loads.

Nonlinear adaptive robust control with continuous error reduction mapping

Abstract Achieving high stability and high-precision control by traditional control algorithms is difficult due to the strong nonlinearity, parameter uncertainty, and random disturbance characteristics of electro-hydraulic proportional servo systems. Herein, we developed a nonlinear adaptive control algorithm based on Lyapunov functions. By estimating the parameters and converging their errors, this algorithm solved the parameter adaptation problem. Additionally, based on convergence inference from the nonnegativity of the derivative of Lyapunov functions, the algorithm designed a disturbance mapping through the derived function. The random disturbance was reduced through the mapping, improving the stability of the system under random disturbance. Finally, we carried out continuous derivation and moved the mapping relationship out of the intermediate temporary controller. This reduced the complexity of the calculation during controller design, prevented performance degradation of the error convergence algorithm after complex calculation, and enhanced the robustness of the algorithm. The algorithm was verified through the simulation model established by Simulink and experience. Results showed that the nonlinear adaptive control algorithm demonstrates fast response, high accuracy, and strong anti-interference ability compared with the high-gain robust control algorithm. It realizes the adaptation of unknown parameters and reduces the control error introduced by random disturbance.

Applications of artificial intelligence in the field of drones

Drones, now equipped with sophisticated sensors and high-resolution cameras, have evolved into highly advanced tools capable of executing a diverse range of tasks with precision and efficiency. The integration of Artificial Intelligence (AI) has been a game-changer, endowing these aerial vehicles with the ability to process information, make autonomous decisions, and perform complex operations without direct human intervention. This paper delves into the transformative potential of AI-integrated drones across various industries, highlighting their role in revolutionizing the way we approach tasks such as autonomous navigation, package delivery, warehouse inventory management, and disaster response. The paper concludes with a comprehensive summary of the findings, emphasizing the potential benefits of AI-integrated drones in various sectors. It also looks forward to future research directions, which will focus on addressing significant challenges such as ensuring data privacy, improving the robustness and reliability of AI algorithms, and navigating the complex regulatory landscape that governs drone usage.

A deep learning framework for predicting endometrial cancer from cytopathologic images with different staining styles.

Endometrial cancer screening is crucial for clinical treatment. Currently, cytopathologists analyze cytopathology images is considered a popular screening method, but manual diagnosis is time-consuming and laborious. Deep learning can provide objective guidance efficiency. But endometrial cytopathology images often come from different medical centers with different staining styles. It decreases the generalization ability of deep learning models in cytopathology images analysis, leading to poor performance. This study presents a robust automated screening framework for endometrial cancer that can be applied to cytopathology images with different staining styles, and provide an objective diagnostic reference for cytopathologists, thus contributing to clinical treatment. We collected and built the XJTU-EC dataset, the first cytopathology dataset that includes segmentation and classification labels. And we propose an efficient two-stage framework for adapting different staining style images, and screening endometrial cancer at the cellular level. Specifically, in the first stage, a novel CM-UNet is utilized to segment cell clumps, with a channel attention (CA) module and a multi-level semantic supervision (MSS) module. It can ignore staining variance and focus on extracting semantic information for segmentation. In the second stage, we propose a robust and effective classification algorithm based on contrastive learning, ECRNet. By momentum-based updating and adding labeled memory banks, it can reduce most of the false negative results. On the XJTU-EC dataset, CM-UNet achieves an excellent segmentation performance, and ECRNet obtains an accuracy of 98.50%, a precision of 99.32% and a sensitivity of 97.67% on the test set, which outperforms other competitive classical models. Our method robustly predicts endometrial cancer on cytopathologic images with different staining styles, which will further advance research in endometrial cancer screening and provide early diagnosis for patients. The code will be available on GitHub.

Comparative analysis of federated learning algorithms under extreme non-IID conditions for computer vision

Data heterogeneity presents enormous challenges for Federated Learning (FL) where data is characterized by non-independent and identically distributed (non-IID) patterns across participating clients. Although a prominent issue, there is a lack of comprehensive studies regarding the efficacy of FL algorithms under different configurations of hyperparameters and models. Three state-of-the-art algorithms were chosen to be the focus of this paper FedAvg, FedProx, and SCAFFOLD. Experiments of the paper study varying degrees of data imbalance and number of local training steps to see their effects on model training results. The algorithms and hyperparameters were evaluated using ResNet50, AlexNet, and DenseNet121 on the CIFAR-10 dataset. Experimental results reveal that FedAvg and SCAFFOLD generally outperform FedProx, except for the latter excelling in the scenario of extreme non-IID data distributions in combination with minimal local training. It was also observed that models with complex architectures like ResNet50 are more susceptible to data imbalances, while simpler models such as AlexNet prove to be more robust. This study provides valuable insights to how different configurations of hyperparameters affect FL in non-IID scenarios, contributing to future development of more efficient and robust FL algorithms.