Instances Of Model Research Articles

Automated Identification of Retinal Disorders by Deep Learning with Optical Coherence Tomography Pictures

A crucial problem in computer-aided diagnosis (CAD) for medical applications is the classification of retinal disorders. This paper addresses the 4-class classification problem in optical coherence tomography image analysis (OCT) to automatically identify choroidal neovascularization (CNV), diabetic macular edema (DME), DRUSEN, and NORMAL. Retinal OCT pictures were identified by the proposed classification approach using a set of four classification model instances, each based on an upgraded residual neural network (ResNet50).At the patient level, the experiment employs a 10-fold cross-validation process that is founded on the growth of the OCT retinal picture collection. According to the study's findings, multiple ResNet50 concatenation is a helpful method when the availability of pictures used in medicine is scarce. In order to better understand the decision-making process, we also ran a bottleneck test and a qualitative assessment of the model predictions.

Dynamic system models and their simulation in the Semantic Web

Modelling and Simulation (M&S) are core tools for designing, analysing and operating today’s industrial systems. They often also represent both a valuable asset and a significant investment. Typically, their use is constrained to a software environment intended to be used by engineers on a single computer. However, the knowledge relevant to a task involving modelling and simulation is in general distributed in nature, even across organizational boundaries, and may be large in volume. Therefore, it is desirable to increase the FAIRness (Findability, Accessibility, Interoperability, and Reuse) of M&S capabilities; to enable their use in loosely coupled systems of systems; and to support their composition and execution by intelligent software agents. In this contribution, the suitability of Semantic Web technologies to achieve these goals is investigated and an open-source proof of concept-implementation based on the Functional Mock-up Interface (FMI) standard is presented. Specifically, models, model instances, and simulation results are exposed through a hypermedia API and an implementation of the Pragmatic Proof Algorithm (PPA) is used to successfully demonstrate the API’s use by a generic software agent. The solution shows an increased degree of FAIRness and fully supports its use in loosely coupled systems. The FAIRness could be further improved by providing more “ rich” (meta)data.

The latent cognitive structures of social networks

Abstract When people are asked to recall their social networks, theoretical and empirical work tells us that they rely on shortcuts, or heuristics. Cognitive social structures (CSSs) are multilayer social networks where each layer corresponds to an individual’s perception of the network. With multiple perceptions of the same network, CSSs contain rich information about how these heuristics manifest, motivating the question, Can we identify people who share the same heuristics? In this work, we propose a method for identifying cognitive structure across multiple network perceptions, analogous to how community detection aims to identify social structure in a network. To simultaneously model the joint latent social and cognitive structure, we study CSSs as three-dimensional tensors, employing low-rank nonnegative Tucker decompositions (NNTuck) to approximate the CSS—a procedure closely related to estimating a multilayer stochastic block model (SBM) from such data. We propose the resulting latent cognitive space as an operationalization of the sociological theory of social cognition by identifying individuals who share relational schema. In addition to modeling cognitively independent, dependent, and redundant networks, we propose a specific model instance and related statistical test for testing when there is social-cognitive agreement in a network: when the social and cognitive structures are equivalent. We use our approach to analyze four different CSSs and give insights into the latent cognitive structures of those networks.

An analysis of information segregation in parallel streams of a multi-stream convolutional neural network.

Visual information is processed in hierarchically organized parallel streams in the primate brain. In the present study, information segregation in parallel streams was examined by constructing a convolutional neural network with parallel architecture in all of the convolutional layers. Although filter weights for convolution were initially set to random values, color information was segregated from shape information in most model instances after training. Deletion of the color-related stream decreased recognition accuracy of animate images, whereas deletion of the shape-related stream decreased recognition accuracy of both animate and inanimate images. The results suggest that properties of filters and functions of a stream are spontaneously segregated in parallel streams of neural networks.

Analysis of chemical reactive nanofluid flow on stretching surface using numerical soft computing approach for thermal enhancement

ABSTRACT In this work, computational intelligence methodologies are used to investigate the trihybrid nanofluid, a new theoretical model with remarkable thermal transmission properties to enhance liquid thermal performance. The nanostructure Cu, Al2O3, and TiO2 were immersed in the base liquid (C2H6O2) to produce (Cu + Al2O3 + TiO2/C2H6O2) trihybrid nanofluid. In a Darcy-Forchheimer porous medium over a stretching Riga sheet, this study examines the electromagnetic ternary hybrid nanofluid flow under various slip situations. The study takes into account the complex interactions between a number of variables, including as viscous dissipation, radiation, heat sources, and chemical reactions. Similarity transformations are used to convert complex partial differential equations for flow, energy, and concentration into nonlinear ordinary differential equations. The highly nonlinear equations of the problem are solved numerically with the use of techniques from the bvp4c approach. The results of the bvp4c method produce the reference dataset required for the Levenberg-Marquardt backpropagation of neural networks (LMBNN). The neural network performance is validated using regression analysis, mean square errors, and error histogram data. The model problem’s consistency and precision are evaluated using the absolute error, which is given for each model instance at around 10−06–10−08, 10−05–10−10 and 10−06 05–10−09. In order to reduce mean square error, the nonlinear fluid dynamics system’s numerical solutions have been taken into consideration. Using the comparative configurations of MSE, error histograms, state transitions, correlation, and regression, the reliability and competence of the stochastic technique are verified.

PARSAC: Accelerating Robust Multi-Model Fitting with Parallel Sample Consensus

We present a real-time method for robust estimation of multiple instances of geometric models from noisy data. Geometric models such as vanishing points, planar homographies or fundamental matrices are essential for 3D scene analysis. Previous approaches discover distinct model instances in an iterative manner, thus limiting their potential for speedup via parallel computation. In contrast, our method detects all model instances independently and in parallel. A neural network segments the input data into clusters representing potential model instances by predicting multiple sets of sample and inlier weights. Using the predicted weights, we determine the model parameters for each potential instance separately in a RANSAC-like fashion. We train the neural network via task-specific loss functions, i.e. we do not require a ground-truth segmentation of the input data. As suitable training data for homography and fundamental matrix fitting is scarce, we additionally present two new synthetic datasets. We demonstrate state-of-the-art performance on these as well as multiple established datasets, with inference times as small as five milliseconds per image.

Automatic Core-Guided Reformulation via Constraint Explanation and Condition Learning

SAT and propagation solvers often underperform for optimisation models whose objective sums many single-variable terms. MaxSAT solvers avoid this by detecting and exploiting cores: subsets of these terms that cannot collectively take their lower bounds. Previous work has shown manual analysis of cores can help define model reformulations likely to speed up solving for many model instances. This paper presents a method to automate this process. For each selected core the method identifies the instance constraints that caused it; infers the model constraints and parameters that explain how these instance constraints were formed; and learns the conditions that made those model constraint instances generate cores, while others did not. It then uses this information to reformulate the objective. The empirical evaluation shows this method can produce useful reformulations. Importantly, the method can be useful in many other situations that require explaining a set of constraints.

Learning to Unlearn: Instance-Wise Unlearning for Pre-trained Classifiers

Since the recent advent of regulations for data protection (e.g., the General Data Protection Regulation), there has been increasing demand in deleting information learned from sensitive data in pre-trained models without retraining from scratch. The inherent vulnerability of neural networks towards adversarial attacks and unfairness also calls for a robust method to remove or correct information in an instance-wise fashion, while retaining the predictive performance across remaining data. To this end, we consider instance-wise unlearning, of which the goal is to delete information on a set of instances from a pre-trained model, by either misclassifying each instance away from its original prediction or relabeling the instance to a different label. We also propose two methods that reduce forgetting on the remaining data: 1) utilizing adversarial examples to overcome forgetting at the representation-level and 2) leveraging weight importance metrics to pinpoint network parameters guilty of propagating unwanted information. Both methods only require the pre-trained model and data instances to forget, allowing painless application to real-life settings where the entire training set is unavailable. Through extensive experimentation on various image classification benchmarks, we show that our approach effectively preserves knowledge of remaining data while unlearning given instances in both single-task and continual unlearning scenarios.

Error analysis of a reduced order method for the Allen-Cahn equation⁎

In this paper we carry out an error analysis for a reduced order method for the Allen-Cahn equation. First, an ensemble of snapshots is formed from the numerical solutions at some time instances of the full order model, which is a time-space discretisation of the Allen-Cahn equation. The reduced order model is essentially a new spatial discretisation method by using low dimensional approximations to the original approximation space. The low dimensional approximation space is generated from the ensemble of snapshots by applying a proper orthogonal decomposition method. To determine the error between the exact solution and the solution of the reduced order model. We consider a time-space discretisation for which an error estimate of the full model solution is available. Specifically, the full discretisation is based on a stabilized auxiliary variable approach for the time stepping and a spectral Galerkin method for the spatial discretisation. The advantages of this full discretisation are its unconditional stability, the availability of error estimates and its ease of implementation. An estimate of the errors in the H1 seminorm is rigorously derived for both the full order model and the reduced order model, which is then verified by some numerical examples.

Międzykulturowa interwencja. Dawid Rogacz rozmawia z Q. Edwardem Wangiem

In this interview conducted at the XXIII International Congress of Historical Sciences in Po-znan, Q. Edward Wang (Professor of History at Rowan University in the US and Changjiang Professor of History at Beijing University in China) addresses the main problems in global and comparative historiography. Elaborating upon the background and impact of his scientific con-tributions, Prof. Wang discusses the modernization of Chinese historical writing, the challenges of comparative historiography, the status of historical studies today, and the future of historical theory. In addition, he refers to his recent studies in culinary history as instances of a new model of global historiography.

DeadWood: Including Disturbance and Decay in the Depiction of Digital Nature

The creation of truly believable simulated natural environments remains an unsolved problem in Computer Graphics. This is, in part, due to a lack of visual variety. In nature, apart from variation due to abiotic and biotic growth factors, a significant role is played by disturbance events, such as fires, windstorms, disease, and death and decay processes, which give rise to both standing dead trees (snags) and downed woody debris (logs). For instance, snags constitute on average 10% of unmanaged forests by basal area, and logs account for 2 \(\frac{1}{2}\) times this quantity. While previous systems have incorporated individual elements of disturbance (e.g., forest fires) and decay (e.g., the formation of humus), there has been no unifying treatment, perhaps because of the challenge of matching simulation results with generated geometric models. In this paper, we present a framework that combines an ecosystem simulation, which explicitly incorporates disturbance events and decay processes, with a model realization process, which balances the uniqueness arising from life history with the need for instancing due to memory constraints. We tested our hypothesis concerning the visual impact of disturbance and decay with a two-alternative forced-choice experiment ( n = 116). Our findings are that the presence of dead wood in various forms, as snags or logs, significantly improves the believability of natural scenes, while, surprisingly, general variation in the number of model instances, with up to 8 models per species, and a focus on disturbance events, does not.

Effect of meniscus modelling assumptions in a static tibiofemoral finite element model: importance of geometry over material.

Finite element studies of the tibiofemoral joint have increased use in research, with attention often placed on the material models. Few studies assess the effect of meniscus modelling assumptions in image-based models on contact mechanics outcomes. This work aimed to assess the effect of modelling assumptions of the meniscus on knee contact mechanics and meniscus kinematics. A sensitivity analysis was performed using three specimen-specific tibiofemoral models and one generic knee model. The assumptions in representing the meniscus attachment on the tibia (shape of the roots and position of the attachment), the material properties of the meniscus, the shape of the meniscus and the alignment of the joint were evaluated, creating 40 model instances. The values of material parameters for the meniscus and the position of the root attachment had a small influence on the total contact area but not on the meniscus displacement or the force balance between condyles. Using 3D shapes to represent the roots instead of springs had a large influence in meniscus displacement but not in knee contact area. Changes in meniscus shape and in knee alignment had a significantly larger influence on all outcomes of interest, with differences two to six times larger than those due to material properties. The sensitivity study demonstrated the importance of meniscus shape and knee alignment on meniscus kinematics and knee contact mechanics, both being more important than the material properties or the position of the roots. It also showed that differences between knees were large, suggesting that clinical interpretations of modelling studies using single geometries should be avoided.

Emulation of Quantitative Systems Pharmacology models to accelerate virtual population inference in immuno-oncology

Quantitative Systems Pharmacology (QSP) models are increasingly being applied for target discovery and dose selection in immuno-oncology (IO). Typical application involves virtual trial, a simulation of a virtual population of hundreds of model instances with model inputs reflecting individual variability. While the structure of the model and initial parameterisation are based on literature describing the underlying biology, calibration of the virtual population by existing clinical data is frequently required to create tumour and patient population specific model instances. Since comparison of a virtual trial with clinical output requires hundreds of large-scale, non-linear model evaluations, the inference of a virtual population is computationally expensive, frequently becoming a bottleneck. Here, we present novel approach to virtual population inference in IO using emulation of the QSP model and an objective function based on Kolmogorov-Smirnov statistics to maximise congruence of simulated and observed clinical tumour size distributions. We sample the parameter space of a QSP IO model to collect a set of tumour growth time profiles. We evaluate performance of several machine learning approaches in interpolating these time profiles and create a surrogate model, which computes tumor growth profiles faster than the original model and allows examination of tens of millions of virtual patients. We use the surrogate model to infer a virtual population maximising congruence with the waterfall plot of a pembrolizumab clinical trial. We believe that our approach is applicable not only in QSP IO, but also in other applications where virtual populations need to be inferred for computationally expensive mechanistic models.

Towards provably efficient quantum algorithms for large-scale machine-learning models

Large machine learning models are revolutionary technologies of artificial intelligence whose bottlenecks include huge computational expenses, power, and time used both in the pre-training and fine-tuning process. In this work, we show that fault-tolerant quantum computing could possibly provide provably efficient resolutions for generic (stochastic) gradient descent algorithms, scaling as O(T2×polylog(n))\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${{{{{{{\\mathcal{O}}}}}}}}({T}^{2}\ imes {{{{{{{\\rm{polylog}}}}}}}}(n))$$\\end{document}, where n is the size of the models and T is the number of iterations in the training, as long as the models are both sufficiently dissipative and sparse, with small learning rates. Based on earlier efficient quantum algorithms for dissipative differential equations, we find and prove that similar algorithms work for (stochastic) gradient descent, the primary algorithm for machine learning. In practice, we benchmark instances of large machine learning models from 7 million to 103 million parameters. We find that, in the context of sparse training, a quantum enhancement is possible at the early stage of learning after model pruning, motivating a sparse parameter download and re-upload scheme. Our work shows solidly that fault-tolerant quantum algorithms could potentially contribute to most state-of-the-art, large-scale machine-learning problems.

PyBEST: Improved functionality and enhanced performance

PyBEST: Improved functionality and enhanced performance

Analysis of Strategies for Enhancing Informatization Teaching Ability of College Teachers Combined with Mathematical Statistics Multiple Regression Models

Abstract In this paper, from the perspective of education informatization, for the problem of teachers’ informatization teaching ability in colleges and universities, the loss function of the analysis of teachers’ informatization teaching ability in colleges and universities can be calculated by using a Logistic regression algorithm. In order to prevent the overfitting problem of multiple regression analysis, it is necessary to optimize the multiple regression analysis by regular transformation, and then construct the model of teachers’ informatization teaching ability based on multiple linear regression, and determine the research variables and regression equations in the model. Starting from the research objectives, the research objects and problems are determined, and two analysis methods, simulation analysis and statistical analysis, are used to empirically analyze the informatization teaching ability of college teachers. The results show that on the analysis of model instances, the correlation coefficients between the factors and the overall teacher educators’ informatization teaching ability range from 0.671 to 0.829, and there is a certain correlation between each factor and teacher educators’ informatization teaching ability. On the analysis of differences in teachers’ informatized teaching competence, the five competence constructs of Guangzhou college teachers’ education in technological literacy (P=0.653>0.05), instructional design (P=0.133>0.05), instructional implementation (P=0.584>0.05), evaluation development (P=0.287>0.05), and informatized teaching competence (P=0.413>0.05) do not significant differences exist. This study comprehensively analyzes the strategies for improving teachers’ informatization teaching ability in order to play some role in promoting the development of education informatization.

On the point process with finite memory and its application to optimal age replacement

There has been extensive study of various repair models in the literature, mostly under the assumption that these repairs are minimal or imperfect/better than minimal. Although this is often a realistic assumption, it may not be sufficient to model instances where the repair is worse than minimal. The generalized Polya process (GPP) that has been used to describe this type of repair takes into account all previous events/repairs, which is not often the case in practice. Therefore, in this paper, we define a new process with finite memory that starts as the GPP but, after a certain number of events or elapsed time, becomes the non-homogeneous Poisson process of repairs (minimal repairs). The corresponding age replacement policy is defined and the optimal solutions that minimize the long-run expected cost rate are analyzed. The detailed numerical examples illustrate our findings.

The Integration of Vocal Music Teaching and Traditional Music Culture in Colleges and Universities in the Context of Internet Plus

Abstract Firstly, the signal extraction in the teaching of vocal music and traditional music in the background of Internet+ is explained in the research. Considering that the initial signal is an analog signal, it is necessary to carry out quantization, pre-emphasis, and windowing and framing processing of the music signal, thus realizing the conversion from analog signals to digital signals. Secondly, the features of the music signal are extracted from the fundamental frequency perturbation and resonance peak perturbation so as to obtain the eigenvalues of the music digital signal. Then, the evaluation model of teaching fusion of vocal music and traditional music culture in colleges and universities is constructed, the input and network parameters of the model are determined according to the music features and variables, and the evaluation model of teaching fusion of vocal music and traditional music based on the Internet is empirically analyzed by means of simulation analysis. The results show that on the evaluation model instance analysis, the average scoring results obtained from the network based on the evaluation model network for the integration of vocal music and traditional music teaching are 0.729, an error of 0.05%, and compared with the subjective scoring of 0.741 (scored by the experts), the two evaluation results are similar, and the output of the evaluation model basically is able to meet the requirements. Through this study, the main content of classroom teaching can be enhanced, and students can be guided to treat traditional music and culture correctly.

Faire: Repairing Fairness of Neural Networks via Neuron Condition Synthesis

Deep Neural Networks (DNNs) have achieved tremendous success in many applications, while it has been demonstrated that DNNs can exhibit some undesirable behaviors on concerns such as robustness, privacy, and other trustworthiness issues. Among them, fairness (i.e., non-discrimination) is one important property, especially when they are applied to some sensitive applications (e.g., finance and employment). However, DNNs easily learn spurious correlations between protected attributes (e.g., age, gender, race) and the classification task and develop discriminatory behaviors if the training data is imbalanced. Such discriminatory decisions in sensitive applications would introduce severe social impacts. To expose potential discrimination problems in DNNs before putting them in use, some testing techniques have been proposed to identify the discriminatory instances (i.e., instances that show defined discrimination 1 ). However, how to repair DNNs after detecting such discrimination is still challenging. Existing techniques mainly rely on retraining on a large number of discriminatory instances generated by testing methods, which requires huge time overhead and makes the repairing inefficient. In this work, we propose the method Faire to effectively and efficiently repair the fairness issues of DNNs, without using additional data (e.g., discriminatory instances). Our basic idea is inspired by the traditional program repair method that synthesizes proper condition checking. To repair traditional programs, a typical method is to localize the program defects and repair the program logic by adding condition checking. Similarly, for DNNs, we try to understand the unfair logic and reformulate it with well-designed condition checking. In this article, we synthesize the condition that can reduce the effect of features relevant to the protected attributes in the DNN. Specifically, we first perform the neuron-based analysis and check the functionalities of neurons to identify neurons whose outputs could be regarded as features relevant to protected attributes and original tasks. Then a new condition layer is added after each hidden layer to penalize neurons that are accountable for the protected features (i.e., intermediate features relevant to protected attributes) and promote neurons that are accountable for the non-protected features (i.e., intermediate features relevant to original tasks). In sum, the repair rate 2 of Faire reaches up to more than 99%, which outperforms other methods, and the whole repairing process only takes no more than 340 s. The evaluation results demonstrate that our approach can effectively and efficiently repair the individual discriminatory instances of the target model.

An extensible schema for capturing environmental model metadata: Implementation in the HydroShare online data repository

Model metadata can be used to document the specifications, inputs, provenance, and other important attributes of an environmental modeling application. An open challenge in creating a general model metadata schema is addressing the diversity of environmental models in a way that provides flexibility and scalability while maintaining a standard core. This paper presents a new model metadata schema design aimed at addressing this challenge. Specifically, we present an extensible metadata schema for describing what we call a model instance: a specific application of an environmental model. We implement the schema in the HydroShare online data repository and demonstrate how the schema can be used for the Soil and Water Assessment (SWAT) model.