A fully supervised universal adversarial perturbations and the progressive optimization

2021 ◽  
pp. 1-10
Author(s):  
Guangling Sun ◽  
Haoqi Hu ◽  
Xinpeng Zhang ◽  
Xiaofeng Lu
Keyword(s):  
Computer Vision ◽  
Supervised Learning ◽  
Optimization Strategy ◽  
Internal Layers ◽  
Latent Factors ◽  
Factors Affecting ◽  
Output Layer ◽  
Depth Analysis ◽  
Adversarial Examples ◽  
Progressive Optimization

Universal Adversarial Perturbations(UAPs), which are image-agnostic adversarial perturbations, have been demonstrated to successfully deceive computer vision models. Proposed UAPs in the case of data-dependent, use the internal layers’ activation or the output layer’s decision values as supervision. In this paper, we use both of them to drive the supervised learning of UAP, termed as fully supervised UAP(FS-UAP), and design a progressive optimization strategy to solve the FS-UAP. Specifically, we define an internal layers supervised objective relying on multiple major internal layers’ activation to estimate the deviations of adversarial examples from legitimate examples. We also define an output layer supervised objective relying on the logits of output layer to evaluate attacking degrees. In addition, we use the UAP found by previous stage as the initial solution of the next stage so as to progressively optimize the UAP stage-wise. We use seven networks and ImageNet dataset to evaluate the proposed FS-UAP, and provide an in-depth analysis for the latent factors affecting the performance of universal attacks. The experimental results show that our FS-UAP (i) has powerful capability of cheating CNNs (ii) has superior transfer-ability across models and weak data-dependent (iii) is appropriate for both untarget and target attacks.

scholarly journals Core Elements Affecting Sharing: Evidence from the United States

2021 ◽  
Vol 13 (7) ◽  
pp. 3943
Author(s):  
Aurelija Burinskienė ◽  
Edita Leonavičienė ◽  
Virginija Grybaitė ◽  
Olga Lingaitienė ◽  
Juozas Merkevičius
Keyword(s):  
United States ◽  
The United States ◽  
Collaborative Consumption ◽  
Factors Affecting ◽  
Use Of Resources ◽  
Personal Consumption ◽  
Depth Analysis ◽  
Core Elements ◽  
Long Time ◽  
Number Of Visits

The new phenomenon called sharing or collaborative consumption emerged a decade ago and is continuously growing. It creates new possibilities for society, and especially for business, is beneficial for the environment, makes more efficient use of resources, and presents a new competitive business model. The scientific literature lacks a more in-depth analysis of the factors influencing sharing activity growth; therefore, the paper’s authors attempt to fill this gap. The authors aim to identify the factors affecting the use of sharing platforms. To reach the goal, the authors developed a regression model and constructed a list of 71 variables. The study used monthly United States data from January 2017 to June 2020 from the publicly available Federal Reserve Economic Data (FRED)and Google trends databases. The comparison to other indexes proves that the proposed index, representing the number of visits to sharing platforms (SEP), is a unique one. The first index allowed us to revise the sharing activity monthly. The authors identified that variables such as wage level, social network users, import level, and personal consumption are critical in affecting the number of visits to sharing platforms. The presented framework could be helpful for practitioners and policymakers analysing the stimulation of sharing or collaborative consumption. It includes indicators representing different areas, such as society, technology, and country, and allows for monthly investigations. Such activity was evident for a long time when online platforms contributed to its wider accessibility. The results help to forecast the number of visits monthly. Sharing is still an emerging area for research; thus, the authors tried to explore the phenomenon of sharing to expand the conceptual level of knowledge.

scholarly journals A Survey on Contrastive Self-Supervised Learning

Technologies ◽  
2020 ◽  
Vol 9 (1) ◽  
pp. 2
Author(s):  
Ashish Jaiswal ◽  
Ashwin Ramesh Babu ◽  
Mohammad Zaki Zadeh ◽  
Debapriya Banerjee ◽  
Fillia Makedon
Keyword(s):  
Computer Vision ◽  
Supervised Learning ◽  
Language Processing ◽  
Large Scale ◽  
Performance Comparison ◽  
Extensive Review ◽  
Future Directions ◽  
Dominant Component ◽  
Supervised Methods ◽  
The Cost

Self-supervised learning has gained popularity because of its ability to avoid the cost of annotating large-scale datasets. It is capable of adopting self-defined pseudolabels as supervision and use the learned representations for several downstream tasks. Specifically, contrastive learning has recently become a dominant component in self-supervised learning for computer vision, natural language processing (NLP), and other domains. It aims at embedding augmented versions of the same sample close to each other while trying to push away embeddings from different samples. This paper provides an extensive review of self-supervised methods that follow the contrastive approach. The work explains commonly used pretext tasks in a contrastive learning setup, followed by different architectures that have been proposed so far. Next, we present a performance comparison of different methods for multiple downstream tasks such as image classification, object detection, and action recognition. Finally, we conclude with the limitations of the current methods and the need for further techniques and future directions to make meaningful progress.

In-Depth Analysis of Key Factors Affecting the Catalysis of Oxidized Carbon Blacks for Cellulose Hydrolysis

ACS Catalysis ◽  
2021 ◽  
pp. 892-905
Author(s):  
Atsushi Gabe ◽  
Akira Takatsuki ◽  
Masahiko Hiratani ◽  
Masato Kaneeda ◽  
Yoshiaki Kurihara ◽  
...  
Keyword(s):  
Cellulose Hydrolysis ◽  
Key Factors ◽  
Factors Affecting ◽  
Carbon Blacks ◽  
Depth Analysis ◽  
Oxidized Carbon

Forecast Model of Phreatic Surface on Tailings Dam Based on GM-GRNN Theory

2010 ◽  
Vol 44-47 ◽  
pp. 3403-3407
Author(s):  
Fei Yue Wang ◽  
Zhi Sheng Xu ◽  
Long Jun Dong
Keyword(s):  
Prediction Method ◽  
Forecast Model ◽  
Random Disturbance ◽  
Tailings Dam ◽  
Phreatic Surface ◽  
Factors Affecting ◽  
Depth Analysis ◽  
Tailing Dam ◽  
Calculation Results ◽  
Surface Prediction

Due to the extremely complicated seepage boundary conditions of tailing dam, the calculation results adopting two-dimensional simplified theory may greatly different from the measured results. It is urgent need of an accurate calculation method to forecast phreatic surface. In-depth analysis of factors affecting tailings dam phreatic surface, phreatic surface prediction model based on GRNN and GM (1,1) was established. A tailing dam engineering is tested using this model. It shows that the model uses the advantages of "accumulative generation" of a Gray prediction method, which weakens the original sequence of random disturbance factors, and increases the regularity of data. It also makes full advantage of the GRNN approximation performance, which has a fast solving speed, describes the nonlinear relationship easily, and avoids the defects of Gray theory.

scholarly journals SLCV–a supervised learning—computer vision combined strategy for automated muscle fibre detection in cross-sectional images

PeerJ ◽  
2019 ◽  
Vol 7 ◽  
pp. e7053
Author(s):  
Anika Rettig ◽  
Tobias Haase ◽  
Alexandr Pletnyov ◽  
Benjamin Kohl ◽  
Wolfgang Ertel ◽  
...  
Keyword(s):  
Computer Vision ◽  
Supervised Learning ◽  
Muscle Fibre ◽  
Cross Sections ◽  
Reconstruction Error ◽  
Muscle Fibres ◽  
Dice Similarity Coefficient ◽  
Cross Sectional ◽  
Staining Methods ◽  
Combined Strategy

Muscle fibre cross-sectional area (CSA) is an important biomedical measure used to determine the structural composition of skeletal muscle, and it is relevant for tackling research questions in many different fields of research. To date, time consuming and tedious manual delineation of muscle fibres is often used to determine the CSA. Few methods are able to automatically detect muscle fibres in muscle fibre cross-sections to quantify CSA due to challenges posed by variation of brightness and noise in the staining images. In this paper, we introduce the supervised learning-computer vision combined pipeline (SLCV), a robust semi-automatic pipeline for muscle fibre detection, which combines supervised learning (SL) with computer vision (CV). SLCV is adaptable to different staining methods and is quickly and intuitively tunable by the user. We are the first to perform an error analysis with respect to cell count and area, based on which we compare SLCV to the best purely CV-based pipeline in order to identify the contribution of SL and CV steps to muscle fibre detection. Our results obtained on 27 fluorescence-stained cross-sectional images of varying staining quality suggest that combining SL and CV performs significantly better than both SL-based and CV-based methods with regards to both the cell separation- and the area reconstruction error. Furthermore, applying SLCV to our test set images yielded fibre detection results of very high quality, with average sensitivity values of 0.93 or higher on different cluster sizes and an average Dice similarity coefficient of 0.9778.

scholarly journals Proposals for university promoting in QS rating based on the methods of statistical analysis

2020 ◽  
Vol 17 (1) ◽  
pp. 35-43
Author(s):  
A. A. Mikryukov ◽  
M. S. Gasparian ◽  
D. S. Karpov
Keyword(s):  
Factor Analysis ◽  
Statistical Analysis ◽  
Performance Indicators ◽  
Factor Model ◽  
Reference Group ◽  
Latent Factors ◽  
University Ranking ◽  
Factors Affecting ◽  
University Performance ◽  
The University

The purpose of the study. The purpose of the study is to develop scientifically based proposals to increase the university performance indicators in the international institutional rating QS to the required values, taking into account the presence of a combination of latent (hidden) factors, the degree of achievement of the set values of the basic indicators and, as a result, the university ranking level.Materials and methods. To achieve this goal, methods of statistical analysis (correlation-regression and factor analysis) were used, which made it possible to identify the degree of influence of latent factors on basic indicators and the main indicator (rating functional). During the study, the following tasks were solved: identification of latent factors affecting the basic indicators of the university, an assessment of their significance and degree of influence on the basic indicators, as well as their grouping. Based on the results of the correlation - regression and factor analysis, measures are formulated to achieve the specified values of the QS University institutional rating indicators.Results. An approach to solving the problem of providing conditions for achieving the required values of university performance indicators in the international institutional ranking QS using models developed based on the methods of correlation-regression and factor analysis is proposed. Estimates of the relationship of indicators and university ranking based on the methods of correlation and regression analysis are obtained. A comparative analysis of the results obtained at the universities of the reference group is made. The problem of identifying factors that influence the change in the values of indicators is solved; the degree of this influence is assessed. Based on the results obtained, reasonable proposals have been developed to achieve the required values of the basic indicators and the rating functional of the university.Conclusion. The results obtained in the course of the study made it possible to justify the measures necessary to solve the problem of achieving the specified performance indicators of the university. Based on the correlation model, correlation dependencies between the rating functional and basic indicators are obtained. Interpretation of the results of factor analysis allowed us to identify a set of factors that have a significant impact on the basic indicators. It is shown that measures to achieve the specified indicators must be carried out taking into account the revealed correlation dependencies between factors and basic indicators, as well as the interpretation results of the developed factor model.

scholarly journals Assessing the caring behaviors of critical care nurses

2018 ◽  
Vol 8 (10) ◽  
pp. 77 ◽  
Author(s):  
Samah Anwar Shalaby ◽  
Nouf Fahad Janbi ◽  
Khairiah Khalid Mohammed ◽  
Kholud Mohammed Al-harthi
Keyword(s):  
Job Satisfaction ◽  
Critical Care ◽  
Educational Background ◽  
Critical Care Nurses ◽  
Affecting Factors ◽  
Caring Behaviors ◽  
Factors Affecting ◽  
Critical Care Units ◽  
Depth Analysis ◽  
The Impact

Objective: To assess the critical care nurses’ perception of their caring behaviors and factors affecting these behaviors.Methods: Participants of this descriptive correlational exploratory study included 277 critical care nurses selected conveniently from nurses worked in all critical care units in King Khalid Hospital, Jeddah. A self-reported questionnaire namely, “Critical Care Nurses Caring Behavior Perception” developed by the researchers after reviewing related literature was used to assess caring behaviors and their affecting factors as perceived by critical care nurses.Results: Seventy percent of the nurses aged between 31 to 50 years old and more than half of nurses had ICU experience ranged from 6 to 10 years, while two thirds of nurses had no previous training about caring behaviors. The study findings revealed that the majority of nurses had high scores of perceived caring behaviors, whereas the mean of their perception was 296.96 ± 18.32. There was a statistical significant positive relationship between nurses’ perception and their work circumstances, workload, job satisfaction, educational background and patient characteristics.Conclusions: It is important to consider critical units’ circumstances, nurses’ educational background, job satisfaction, as well as the nature of critically ill patients in order to promote nurses awareness and implementation of caring behaviors. Moreover, replication of the current study using qualitative approach for in-depth analysis of the impact of factors could affecting caring behaviors on nurses’ perception in various highly specialized critical care units.

Evaluation of innovative development of housing construction in the partition of federal districts

2021 ◽  
pp. 199-223
Author(s):  
Ekaterina Litvinenko
Keyword(s):  
Construction Industry ◽  
Housing Construction ◽  
Purchasing Power ◽  
Innovative Development ◽  
Factors Affecting ◽  
Depth Analysis ◽  
Federal Districts ◽  
The Russian Federation ◽  
Innovative Potential ◽  
North Western

The aim of the study is to assess the innovative potential of housing construction through determining the purchasing power of citizens acquiring housing in the property. In terms of hypothesis the study offers the term of acquiring housing in property as a criterion for this assessment. This indicator displays the change in the purchasing power of citizens under the influence of various externalities. The analysis of this indicator allows us to determine the influence of these factors on the change in the purchasing power of population, to identify the reserves that contribute to its increase, and to identify the policies aimed at innovative development of construction industry. The article assesses the innovative development of housing through an in-depth analysis of factors affecting the change in the term of housing acquisition in the Russian Federation at large, as well as in the context of individual subjects of the Central and North-Western Federal Districts of Russia in the period from 2017 to 2018.

scholarly journals Robust Semi-Supervised Traffic Sign Recognition via Self-Training and Weakly-Supervised Learning

Sensors ◽  
2020 ◽  
Vol 20 (9) ◽  
pp. 2684 ◽  
Author(s):  
Obed Tettey Nartey ◽  
Guowu Yang ◽  
Sarpong Kwadwo Asare ◽  
Jinzhao Wu ◽  
Lady Nadia Frempong
Keyword(s):  
Machine Learning ◽  
Computer Vision ◽  
Supervised Learning ◽  
Machine Learning Algorithms ◽  
Unbalanced Data ◽  
Traffic Sign Recognition ◽  
Training Set ◽  
Traffic Sign ◽  
Sign Recognition ◽  
Weakly Supervised

Traffic sign recognition is a classification problem that poses challenges for computer vision and machine learning algorithms. Although both computer vision and machine learning techniques have constantly been improved to solve this problem, the sudden rise in the number of unlabeled traffic signs has become even more challenging. Large data collation and labeling are tedious and expensive tasks that demand much time, expert knowledge, and fiscal resources to satisfy the hunger of deep neural networks. Aside from that, the problem of having unbalanced data also poses a greater challenge to computer vision and machine learning algorithms to achieve better performance. These problems raise the need to develop algorithms that can fully exploit a large amount of unlabeled data, use a small amount of labeled samples, and be robust to data imbalance to build an efficient and high-quality classifier. In this work, we propose a novel semi-supervised classification technique that is robust to small and unbalanced data. The framework integrates weakly-supervised learning and self-training with self-paced learning to generate attention maps to augment the training set and utilizes a novel pseudo-label generation and selection algorithm to generate and select pseudo-labeled samples. The method improves the performance by: (1) normalizing the class-wise confidence levels to prevent the model from ignoring hard-to-learn samples, thereby solving the imbalanced data problem; (2) jointly learning a model and optimizing pseudo-labels generated on unlabeled data; and (3) enlarging the training set to satisfy the hunger of deep learning models. Extensive evaluations on two public traffic sign recognition datasets demonstrate the effectiveness of the proposed technique and provide a potential solution for practical applications.

scholarly journals Human-Like Sketch Object Recognition via Analogical Learning

2019 ◽  
Vol 33 ◽  
pp. 1336-1343
Author(s):  
Kezhen Chen ◽  
Irina Rabkina ◽  
Matthew D. McLure ◽  
Kenneth D. Forbus
Keyword(s):  
Computer Vision ◽  
Deep Learning ◽  
Object Recognition ◽  
Image Recognition ◽  
Visual Representations ◽  
Learning Systems ◽  
Training Data ◽  
Adversarial Examples ◽  
Analogical Learning

Deep learning systems can perform well on some image recognition tasks. However, they have serious limitations, including requiring far more training data than humans do and being fooled by adversarial examples. By contrast, analogical learning over relational representations tends to be far more data-efficient, requiring only human-like amounts of training data. This paper introduces an approach that combines automatically constructed qualitative visual representations with analogical learning to tackle a hard computer vision problem, object recognition from sketches. Results from the MNIST dataset and a novel dataset, the Coloring Book Objects dataset, are provided. Comparison to existing approaches indicates that analogical generalization can be used to identify sketched objects from these datasets with several orders of magnitude fewer examples than deep learning systems require.

Sign in / Sign up

Export Citation Format

Share Document