scholarly journals SEMANTIC BORROWING FOR GENERALIZED ZERO-SHOT LEARNING

2021 ◽  
Author(s):  
Xiao-wei CHEN
Keyword(s):  
Metric Learning ◽  
Training Process ◽  
Training Set ◽  
Shot Classification ◽  
Testing Data ◽  
Benchmark Datasets ◽  
Novel Method ◽  
Testing Semantics ◽  
The Relationship ◽  
Generalized Zero

<p>Generalized zero-shot learning (GZSL) is one of the most realistic problems, but also one of the most challenging problems due to the partiality of the classifier to supervised classes. Instance-borrowing methods and synthesizing methods solve this problem to some extent with the help of testing semantics, but therefore neither can be used under the class-inductive instance-inductive (CIII) training setting where testing data are not available, and the latter require the training process of a classifier after generating examples. In contrast, a novel method called Semantic Borrowing for improving GZSL methods with compatibility metric learning under CIII is proposed in this paper. It borrows similar semantics in the training set, so that the classifier can model the relationship between the semantics of zero-shot and supervised classes more accurately during training. In practice, the information of semantics of unseen or unknown classes would not be available for training while this approach does NOT need any information of semantics of unseen or unknown classes. The experimental results on representative GZSL benchmark datasets show that it can reduce the partiality of the classifier to supervised classes and improve the performance of generalized zero-shot classification.</p>

scholarly journals SEMANTIC BORROWING FOR GENERALIZED ZERO-SHOT LEARNING

2021 ◽  
Author(s):  
Xiao-wei CHEN
Keyword(s):  
Metric Learning ◽  
Training Process ◽  
Training Set ◽  
Shot Classification ◽  
Testing Data ◽  
Benchmark Datasets ◽  
Novel Method ◽  
Testing Semantics ◽  
The Relationship ◽  
Generalized Zero

<p>Generalized zero-shot learning (GZSL) is one of the most realistic problems, but also one of the most challenging problems due to the partiality of the classifier to supervised classes. Instance-borrowing methods and synthesizing methods solve this problem to some extent with the help of testing semantics, but therefore neither can be used under the class-inductive instance-inductive (CIII) training setting where testing data are not available, and the latter require the training process of a classifier after generating examples. In contrast, a novel method called Semantic Borrowing for improving GZSL methods with compatibility metric learning under CIII is proposed in this paper. It borrows similar semantics in the training set, so that the classifier can model the relationship between the semantics of zero-shot and supervised classes more accurately during training. In practice, the information of semantics of unseen or unknown classes would not be available for training while this approach does NOT need any information of semantics of unseen or unknown classes. The experimental results on representative GZSL benchmark datasets show that it can reduce the partiality of the classifier to supervised classes and improve the performance of generalized zero-shot classification.</p>

Zero-shot Metric Learning

2019 ◽  
Author(s):  
Xinyi Xu ◽  
Huanhuan Cao ◽  
Yanhua Yang ◽  
Erkun Yang ◽  
Cheng Deng
Keyword(s):  
State Of The Art ◽  
Metric Learning ◽  
Visual Similarity ◽  
Distance Metric ◽  
Learning Problem ◽  
Combine Data ◽  
Benchmark Datasets ◽  
Novel Method ◽  
Multiple Relation ◽  
Continuous Relation

In this work, we tackle the zero-shot metric learning problem and propose a novel method abbreviated as ZSML, with the purpose to learn a distance metric that measures the similarity of unseen categories (even unseen datasets). ZSML achieves strong transferability by capturing multi-nonlinear yet continuous relation among data. It is motivated by two facts: 1) relations can be essentially described from various perspectives; and 2) traditional binary supervision is insufficient to represent continuous visual similarity. Specifically, we first reformulate a collection of specific-shaped convolutional kernels to combine data pairs and generate multiple relation vectors. Furthermore, we design a new cross-update regression loss to discover continuous similarity. Extensive experiments including intra-dataset transfer and inter-dataset transfer on four benchmark datasets demonstrate that ZSML can achieve state-of-the-art performance.

scholarly journals Global Random Graph Convolution Network for Hyperspectral Image Classification

2021 ◽  
Vol 13 (12) ◽  
pp. 2285
Author(s):  
Chaozi Zhang ◽  
Jianli Wang ◽  
Kainan Yao
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Random Graph ◽  
Adjacency Matrix ◽  
Hyperspectral Image ◽  
Learning Methods ◽  
Current State ◽  
Benchmark Datasets ◽  
Novel Method ◽  
The Relationship

Machine learning and deep learning methods have been employed in the hyperspectral image (HSI) classification field. Of deep learning methods, convolution neural network (CNN) has been widely used and achieved promising results. However, CNN has its limitations in modeling sample relations. Graph convolution network (GCN) has been introduced to HSI classification due to its demonstrated ability in processing sample relations. Introducing GCN into HSI classification, the key issue is how to transform HSI, a typical euclidean data, into non-euclidean data. To address this problem, we propose a supervised framework called the Global Random Graph Convolution Network (GR-GCN). A novel method of constructing the graph is adopted for the network, where the graph is built by randomly sampling from the labeled data of each class. Using this technique, the size of the constructed graph is small, which can save computing resources, and we can obtain an enormous quantity of graphs, which also solves the problem of insufficient samples. Besides, the random combination of samples can make the generated graph more diverse and make the network more robust. We also use a neural network with trainable parameters, instead of artificial rules, to determine the adjacency matrix. An adjacency matrix obtained by a neural network is more flexible and stable, and it can better represent the relationship between nodes in a graph. We perform experiments on three benchmark datasets, and the results demonstrate that the GR-GCN performance is competitive with that of current state-of-the-art methods.

scholarly journals Tree-weighting for multi-study ensemble learners

2019 ◽  
Author(s):  
Maya Ramchandran ◽  
Prasad Patil ◽  
Giovanni Parmigiani
Keyword(s):  
Random Forest ◽  
Random Forests ◽  
Training Process ◽  
Training Set ◽  
Prediction Ability ◽  
True Outcome ◽  
The Individual ◽  
Individual Trees ◽  
The Relationship ◽  
Shed Light

Multi-study learning uses multiple training studies, separately trains classifiers on individual studies, and then forms ensembles with weights rewarding members with better cross-study prediction ability. This article considers novel weighting approaches for constructing tree-based ensemble learners in this setting. Using Random Forests as a single-study learner, we perform a comparison of either weighting each forest to form the ensemble, or extracting the individual trees trained by each Random Forest and weighting them directly. We consider weighting approaches that reward cross-study replicability within the training set. We find that incorporating multiple layers of ensembling in the training process increases the robustness of the resulting predictor. Furthermore, we explore the mechanisms by which the ensembling weights correspond to the internal structure of trees to shed light on the important features in determining the relationship between the Random Forests algorithm and the true outcome model. Finally, we apply our approach to genomic datasets and show that our method improves upon the basic multi-study learning paradigm.

3D-QSAR and Docking Studies on Pyrimidine Derivatives as CSF-1R Inhibitors

2020 ◽  
Vol 17 (3) ◽  
pp. 341-355
Author(s):  
Ya-ting Deng ◽  
Jun-wei Wang ◽  
Han Chu ◽  
Juan Wang ◽  
Yong Hu ◽  
...  
Keyword(s):  
Poor Prognosis ◽  
3D Qsar ◽  
Docking Studies ◽  
Colony Stimulating Factor ◽  
Training Set ◽  
Pyrimidine Derivatives ◽  
Comfa Model ◽  
Inhibitory Activities ◽  
The Relationship ◽  
Stimulating Factor

Background: Colony Stimulating Factor-1 Receptor (CSF-1R) is associated with malignancy, invasiveness and poor prognosis of tumors, and pyrimidine derivatives are considered as a novel class of CSF-1R inhibitor. Methods: To explore the relationship between the structures of substituted pyrimidine derivatives and their inhibitory activities against CSF-1R, CoMFA and CoMSIA analyses, and molecular docking studies were performed on a dataset of forty-four compounds. Results: We found in CoMFA model including steric and electrostatic fields for the training set, the cross-validated q2 value was 0.617 and the non-cross-validated r2 value was 0.983. While, the crossvalidated q2 value was 0.637 and the non-cross-validated r2 value was 0.984 in CoMSIA Model which include steric, electrostatic and hydrophobic fields. 3D equipotential maps generated from CoMFA and CoMSIA along with the docking binding structures provided enough information about the structural requirements for better activity. Conclusion: The data generated from the present study helped us to predict the activity of new inhibitors and further design some novel and potent CSF-1R inhibitors.

scholarly journals A new diatom training set for the reconstruction of past water pH in the Tatra Mountain lakes

2021 ◽  
Author(s):  
Elwira Sienkiewicz ◽  
Michał Gąsiorowski ◽  
Ladislav Hamerlík ◽  
Peter Bitušík ◽  
Joanna Stańczak
Keyword(s):  
Surface Sediments ◽  
Tatra Mountains ◽  
Training Set ◽  
Statistical Parameters ◽  
The Past ◽  
Water Ph ◽  
Inferred Ph ◽  
The 1960S ◽  
The Tatra Mountains ◽  
The Relationship

AbstractLakes located in the Polish and Slovak parts of the Tatra Mountains were included in the Tatra diatom database (POL_SLOV training set). The relationship between the diatoms and the water chemistry in the surface sediments of 33 lakes was the basis for the statistical and numerical techniques for quantitative pH reconstruction. The reconstruction of the past water pH was performed using the alpine (AL:PE) and POL_SLOV training sets to compare the reliability of the databases for the Tatra lakes. The results showed that the POL_SLOV training set had better statistical parameters (R2 higher by 0.16, RMSE and max. bias lower by 0.2 and 0.36, respectively) compared to the AL:PE training set. The better performance of the POL_SLOV training set is particularly visible in the case of Przedni Staw Polski where the curve of the inferred water pH shows an opposite trend for the period from the 1960s to 1990 compared to that based on the AL:PE dataset. The reliability of the inferred pH was confirmed by the comparison with current instrumental measurements.

scholarly journals Dynamics of Fourier Modes in Torus Generative Adversarial Networks

Mathematics ◽  
2021 ◽  
Vol 9 (4) ◽  
pp. 325
Author(s):  
Ángel González-Prieto ◽  
Alberto Mozo ◽  
Edgar Talavera ◽  
Sandra Gómez-Canaval
Keyword(s):  
Fourier Series ◽  
Generative Adversarial Networks ◽  
Learning Models ◽  
Training Process ◽  
Small Perturbations ◽  
Adversarial Networks ◽  
Novel Method ◽  
Truncated Fourier Series ◽  
Real Flow ◽  
Machine Learning Models

Generative Adversarial Networks (GANs) are powerful machine learning models capable of generating fully synthetic samples of a desired phenomenon with a high resolution. Despite their success, the training process of a GAN is highly unstable, and typically, it is necessary to implement several accessory heuristics to the networks to reach acceptable convergence of the model. In this paper, we introduce a novel method to analyze the convergence and stability in the training of generative adversarial networks. For this purpose, we propose to decompose the objective function of the adversary min–max game defining a periodic GAN into its Fourier series. By studying the dynamics of the truncated Fourier series for the continuous alternating gradient descend algorithm, we are able to approximate the real flow and to identify the main features of the convergence of GAN. This approach is confirmed empirically by studying the training flow in a 2-parametric GAN, aiming to generate an unknown exponential distribution. As a by-product, we show that convergent orbits in GANs are small perturbations of periodic orbits so the Nash equillibria are spiral attractors. This theoretically justifies the slow and unstable training observed in GANs.

scholarly journals Mechanics of the small punch test: a review and qualification of additive manufacturing materials

2021 ◽  
Vol 56 (18) ◽  
pp. 10707-10744
Author(s):  
Jonathan Torres ◽  
Ali P. Gordon
Keyword(s):  
Additive Manufacturing ◽  
Material Properties ◽  
Materials Science ◽  
Contemporary Literature ◽  
The Body ◽  
Small Punch Test ◽  
Small Punch ◽  
Punch Test ◽  
Testing Data ◽  
The Relationship

AbstractThe small punch test (SPT) was developed for situations where source material is scarce, costly or otherwise difficult to acquire, and has been used for assessing components with variable, location-dependent material properties. Although lacking standardization, the SPT has been employed to assess material properties and verified using traditional testing. Several methods exist for equating SPT results with traditional stress–strain data. There are, however, areas of weakness, such as fracture and fatigue approaches. This document outlines the history and methodologies of SPT, reviewing the body of contemporary literature and presenting relevant findings and formulations for correlating SPT results with conventional tests. Analysis of literature is extended to evaluating the suitability of the SPT for use with additively manufactured (AM) materials. The suitability of this approach is shown through a parametric study using an approximation of the SPT via FEA, varying material properties as would be seen with varying AM process parameters. Equations describing the relationship between SPT results and conventional testing data are presented. Correlation constants dictating these relationships are determined using an accumulation of data from the literature reviewed here, along with novel experimental data. This includes AM materials to assess the fit of these and provide context for a wider view of the methodology and its interest to materials science and additive manufacturing. A case is made for the continued development of the small punch test, identifying strengths and knowledge gaps, showing need for standardization of this simple yet highly versatile method for expediting studies of material properties and optimization.

Criminal Liability of Athletes for Damage Caused as a Result of Violation of the Rules of Sports Events

2020 ◽  
Vol 12 ◽  
pp. 100-105
Author(s):  
D. A. Bezborodov ◽  
◽  
R. M. Kravchenko ◽  
Keyword(s):  
Law Enforcement ◽  
Sports Injuries ◽  
Criminal Liability ◽  
Sporting Events ◽  
Sports Training ◽  
Training Process ◽  
Sports Organizations ◽  
Sports Events ◽  
Special Rules ◽  
The Relationship

The article deals with issues related to the characteristics of the qualification of causing injury or death to an athlete during sports events. The article analyzes the possibility of applying the provisions of certain circumstances that exclude the criminality of the act. Take into account that the relationship between the participants of sports competitions and sports training, while relationships at the same time are not regulated by the law and sports regulations sports, and the internal rules of sports organizations, defining the organization of the training process. Therefore, the issues related to the influence of special rules regulating the procedure for conducting sports competitions and other sporting events on the features of criminal liability (in particular, guilt), both athletes and other persons who ensure the conduct of sports events, are studied specifically. It is taken into account that modern legislation and law enforcement often ignores this requirement, which, in particular, is expressed in the failure to include the facts of sports injuries in the list of crimes in the field of sports. First of all, the article analyzes the issues of criminal-legal assessment of an athlete's act in the event of injury to health or death to another athlete, given that in sports, harm is usually caused unintentionally, by negligence. Therefore, the work analyzes the risks, harm to health, as well as measures that should have been taken by the organizers of the competition to avoid causing harm, taking into account that all these issues are evaluative. The characteristic of harming an athlete while observing the rules of events by his opponent is given. The question of how the rules relating to a particular sport can exempt a person from liability for causing harm is being investigated.

scholarly journals An Empirical Investigation on a Multiple Filters-Based Approach for Remaining Useful Life Prediction

Machines ◽  
2018 ◽  
Vol 6 (3) ◽  
pp. 35 ◽  
Author(s):  
Hung-Cuong Trinh ◽  
Yung-Keun Kwon
Keyword(s):  
Scoring Function ◽  
Principal Component ◽  
Remaining Useful Life ◽  
Filter Method ◽  
Feature Construction ◽  
Training Set ◽  
Robust Solution ◽  
Benchmark Datasets ◽  
Useful Life ◽  
Better Than

Feature construction is critical in data-driven remaining useful life (RUL) prediction of machinery systems, and most previous studies have attempted to find a best single-filter method. However, there is no best single filter that is appropriate for all machinery systems. In this work, we devise a straightforward but efficient approach for RUL prediction by combining multiple filters and then reducing the dimension through principal component analysis. We apply multilayer perceptron and random forest methods to learn the underlying model. We compare our approach with traditional single-filtering approaches using two benchmark datasets. The former approach is significantly better than the latter in terms of a scoring function with a penalty for late prediction. In particular, we note that selecting a best single filter over the training set is not efficient because of overfitting. Taken together, we validate that our multiple filters-based approach can be a robust solution for RUL prediction of various machinery systems.

Sign in / Sign up

Export Citation Format

Share Document