Encapsulating the Impact of Transfer Learning, Domain Knowledge and Training Strategies in Deep-Learning Based Architecture: A Biometric Based Case Study

With the exponential growth of high-quality fake images in social networks and media, it is necessary to develop recognition algorithms for this type of content. One of the most common types of image and video editing consists of duplicating areas of the image, known as the copy-move technique. Traditional image processing approaches manually look for patterns related to the duplicated content, limiting their use in mass data classification. In contrast, approaches based on deep learning have shown better performance and promising results, but they present generalization problems with a high dependence on training data and the need for appropriate selection of hyperparameters. To overcome this, we propose two approaches that use deep learning, a model by a custom architecture and a model by transfer learning. In each case, the impact of the depth of the network is analyzed in terms of precision (P), recall (R) and F1 score. Additionally, the problem of generalization is addressed with images from eight different open access datasets. Finally, the models are compared in terms of evaluation metrics, and training and inference times. The model by transfer learning of VGG-16 achieves metrics about 10% higher than the model by a custom architecture, however, it requires approximately twice as much inference time as the latter.

Download Full-text

The Ecuadorian Banana Farms Managers’ Perceptions: Innovation as a Driver of Environmental Sustainability Practices

Agriculture ◽

10.3390/agriculture11030213 ◽

2021 ◽

Vol 11 (3) ◽

pp. 213

Author(s):

Alicia Ramírez-Orellana ◽

Daniel Ruiz-Palomo ◽

Alfonso Rojo-Ramírez ◽

John E. Burgos-Burgos

Keyword(s):

Information Systems ◽

Environmental Sustainability ◽

Structural Equation ◽

Equation Modeling ◽

Environmental Practices ◽

Sustainability Practices ◽

Policy Incentives ◽

The Impact ◽

And Training

This article aims to explore the perceptions of banana farms managers towards environmental sustainability practices through the impact of innovation, adoption of information systems, and training employees through a case study in the province of El Oro (Ecuador). Furthermore, the paper assesses how farmers’ perceptions could guide public policy incentives. PLS-Structural Equation Modeling are used as the framework by which the constructs is represented within the model. The model explained 59% of the environmental sustainability practices of Ecuadorian banana farms. The results indicate that environmental sustainability practices were positively influenced mainly by training employees, innovation, and adoption of information systems. Additionally, both the adoption of information systems and training employees indirectly influenced sustainable practices through innovation as a mediator. We may conclude that in the Ecuadorian banana farms, changes in environmental practices are derived from innovation strategies as an axis of development of useful information and training employees in public policies.

Download Full-text

Professional Identity

10.3366/edinburgh/9781474424929.003.0011 ◽

2020 ◽

pp. 227-246

Author(s):

Aaron Ackerley

Keyword(s):

Professional Identity ◽

Cultural Changes ◽

Newspaper Industry ◽

Fourth Estate ◽

Wide Range ◽

Key Topics ◽

British Press ◽

The Impact ◽

And Training

This chapter surveys changing notions of professional identity in the twentieth-century British press. The term ‘journalist’ is highly contested, covering a wide range of figures with different forms of experience and training as well as a wide range of roles within and beyond news organisations. Journalism has also lacked the clearly defined rules of practice and established pathways into the occupation evident in other careers that are classed as professions, such as medicine and law. By exploring key topics such as continuities from the nineteenth-century press, the rise of professionalism and journalists’ associations and unions, the myth of the ‘Fourth Estate’ and struggles over press regulation, and the impact of digitisation, this chapter explains how notions of professional identity within journalism have changed in response to wider social and cultural changes and changes within the newspaper industry itself. These topics are also explored in short case study, focused on the Guardian.

Download Full-text

Conditional factors for training activities in Chinese, Indian and Mexican subsidiaries of German companies

European Journal of Training and Development ◽

10.1108/ejtd-04-2020-0066 ◽

2020 ◽

Vol ahead-of-print (ahead-of-print) ◽

Author(s):

Beke Vogelsang ◽

Matthias Pilz

Keyword(s):

Influencing Factors ◽

External Factors ◽

Further Training ◽

Host Countries ◽

Foreign Investments ◽

Content Type ◽

Training Strategies ◽

The Impact ◽

And Training ◽

National Borders

Purpose The purpose of this paper is to analyse the qualification measures of 12 German multi-national companies (MNCs), all of which are present in China, India and Mexico. In particular, the transfer of dual initial training practices and further training measures are investigated. It examines the impact consistent training strategies across national borders have emerged in German companies or local arrangements have developed despite identical internal influencing factors. Design/methodology/approach Because of its design, the focus is on the external factors that influence the companies’ training measures. However, an exploratory approach was followed. To pursue the research question face-to-face expert interviews were conducted with 46 training managers in 12 active companies in all 3 countries. The interviews were completely transcribed and evaluated using qualitative methods. Findings The analysis shows that it is not internal company factors but country-specific contextual factors that influence training measures and that companies cannot act in the same way worldwide. Research limitations/implications The study is based on 12 MNC and only analyses the blue-collar area. Therefore, it would have to be evaluated whether a similar analysis would result from a survey of other companies in different sectors or whether the differences in terms of training and further training measures would then be even greater. Practical implications The study supports the internationalization strategies of MNC by providing first-hand empirical results concerning recruitment and training of blue colour workers on an intermediate skill level. It gives evidence on the need of national adaptation in the process of transferring training cultures from countries of origin into the host countries. More attention must, therefore, be paid to external factors when developing and implementing training measures. Social implications The economic development in many countries includes an expansion of foreign investments. MNC provides employment and income for workers and their families. However, successful foreign investments also include sustainable recruitment and training strategies of the local workforce. The results of the study support policymakers to guide and support foreign companies to develop successful Human Resource Management strategies in the host countries. Originality/value This paper is original because due to the research design the internal factors are kept largely constant and the external influencing factors are singularly focused in detail. Therefore, this procedure makes it possible to investigate whether consistency training strategies across national borders have emerged in German companies or local arrangements have developed despite identical internal influencing factors.

Download Full-text

Marrying Medical Domain Knowledge With Deep Learning on Electronic Health Records: A Deep Visual Analytics Approach (Preprint)

10.2196/preprints.20645 ◽

2020 ◽

Author(s):

Rui Li ◽

Changchang Yin ◽

Samuel Yang ◽

Buyue Qian ◽

Ping Zhang

Keyword(s):

Heart Failure ◽

Deep Learning ◽

Risk Prediction ◽

Visual Analytics ◽

Domain Knowledge ◽

Knowledge Graph ◽

Data Set ◽

Medical Domain ◽

Clinical Risks

BACKGROUND Deep learning models have attracted significant interest from health care researchers during the last few decades. There have been many studies that apply deep learning to medical applications and achieve promising results. However, there are three limitations to the existing models: (1) most clinicians are unable to interpret the results from the existing models, (2) existing models cannot incorporate complicated medical domain knowledge (eg, a disease causes another disease), and (3) most existing models lack visual exploration and interaction. Both the electronic health record (EHR) data set and the deep model results are complex and abstract, which impedes clinicians from exploring and communicating with the model directly. OBJECTIVE The objective of this study is to develop an interpretable and accurate risk prediction model as well as an interactive clinical prediction system to support EHR data exploration, knowledge graph demonstration, and model interpretation. METHODS A domain-knowledge–guided recurrent neural network (DG-RNN) model is proposed to predict clinical risks. The model takes medical event sequences as input and incorporates medical domain knowledge by attending to a subgraph of the whole medical knowledge graph. A global pooling operation and a fully connected layer are used to output the clinical outcomes. The middle results and the parameters of the fully connected layer are helpful in identifying which medical events cause clinical risks. DG-Viz is also designed to support EHR data exploration, knowledge graph demonstration, and model interpretation. RESULTS We conducted both risk prediction experiments and a case study on a real-world data set. A total of 554 patients with heart failure and 1662 control patients without heart failure were selected from the data set. The experimental results show that the proposed DG-RNN outperforms the state-of-the-art approaches by approximately 1.5%. The case study demonstrates how our medical physician collaborator can effectively explore the data and interpret the prediction results using DG-Viz. CONCLUSIONS In this study, we present DG-Viz, an interactive clinical prediction system, which brings together the power of deep learning (ie, a DG-RNN–based model) and visual analytics to predict clinical risks and visually interpret the EHR prediction results. Experimental results and a case study on heart failure risk prediction tasks demonstrate the effectiveness and usefulness of the DG-Viz system. This study will pave the way for interactive, interpretable, and accurate clinical risk predictions.

Download Full-text

Dropout, a basic and effective regularization method for a deep learning model: a case study

Indonesian Journal of Electrical Engineering and Computer Science ◽

10.11591/ijeecs.v24.i2.pp1009-1016 ◽

2021 ◽

Vol 24 (2) ◽

pp. 1009

Author(s):

Brahim Jabir ◽

Noureddine Falih

Keyword(s):

Deep Learning ◽

Weed Species ◽

It Impacts ◽

Artificial Neurons ◽

Pest Detection ◽

Learning Set ◽

Validation Set ◽

Deep Learning Model ◽

And Training

Deep learning is based on a network of artificial neurons inspired by the human brain. This network is made up of tens or even hundreds of "layers" of neurons. The fields of application of deep learning are indeed multiple; Agriculture is one of those fields in which deep learning is used in various agricultural problems (disease detection, pest detection, and weed identification). A major problem with deep learning is how to create a model that works well, not only on the learning set but also on the validation set. Many approaches used in neural networks are explicitly designed to reduce overfit, possibly at the expense of increasing validation accuracy and training accuracy. In this paper, a basic technique (dropout) is proposed to minimize overfit, we integrated it into a convolutional neural network model to classify weed species and see how it impacts performance, a complementary solution (exponential linear units) are proposed to optimize the obtained results. The results showed that these proposed solutions are practical and highly accurate, enabling us to adopt them in deep learning models.

Download Full-text

The Role of Organization Culture, Leadership and Training Towards Improving Work Performance Country Civil Worker: Case Study in Financial and Development Supervisory Agency

International Journal of Business Studies ◽

10.32924/ijbs.v4i2.143 ◽

2020 ◽

Vol 4 (2) ◽

pp. 115-126

Author(s):

Muhammad Ramaditya ◽

Amirul Wahid Prihantoro

Keyword(s):

Organizational Culture ◽

Random Sampling ◽

Work Performance ◽

Sampling Technique ◽

Simple Random Sampling ◽

Organization Culture ◽

The Impact ◽

And Training

The objective of this study is analyzed the impact of organizational culture, training, and leadership on the work performance of civil workers in financial and development supervisory agency. This research uses an associative approach with simple random sampling technique which is measured by SEM using the Smart PLS 3.0 application. The population of this study was 330 civil workers in Financial and Development Supervisory Agency (BPKP). The results of the study shown that the Organizational Culture Variables did not significantly influence work Performance of the civil worker in Financial and Development Supervisory Agency, but do not pass the reliability test. Training Variables have a positive and significant effect on the work performance of the civil workers. Leadership variables have a positive and significant effect on the work performance of the civil workers in Financial and Development Supervisory Agency.

Download Full-text

Efficient and effective OCR engine training

International Journal on Document Analysis and Recognition (IJDAR) ◽

10.1007/s10032-019-00347-8 ◽

2019 ◽

Vol 23 (1) ◽

pp. 73-88

Author(s):

Christian Clausner ◽

Apostolos Antonacopoulos ◽

Stefan Pletschacher

Keyword(s):

Evaluation System ◽

Training Data ◽

Text Recognition ◽

Training Approach ◽

Efficiency And Effectiveness ◽

Analysis System ◽

The Impact ◽

Data Requirements ◽

And Training

Abstract We present an efficient and effective approach to train OCR engines using the Aletheia document analysis system. All components required for training are seamlessly integrated into Aletheia: training data preparation, the OCR engine’s training processes themselves, text recognition, and quantitative evaluation of the trained engine. Such a comprehensive training and evaluation system, guided through a GUI, allows for iterative incremental training to achieve best results. The widely used Tesseract OCR engine is used as a case study to demonstrate the efficiency and effectiveness of the proposed approach. Experimental results are presented validating the training approach with two different historical datasets, representative of recent significant digitisation projects. The impact of different training strategies and training data requirements is presented in detail.

Download Full-text

TripAdvisor and reputation: a case study of the hotel industry in Cyprus

EuroMed Journal of Business ◽

10.1108/emjb-11-2016-0031 ◽

2017 ◽

Vol 12 (3) ◽

pp. 316-334 ◽

Cited By ~ 13

Author(s):

Nicholas Nicoli ◽

Evgenia Papadopoulou

Keyword(s):

Social Media ◽

Hotel Industry ◽

Communication Strategy ◽

Reputation Management ◽

Annual Reports ◽

Content Type ◽

Online Reputation ◽

The Impact ◽

And Training

Purpose The purpose of this paper is to examine the significance of TripAdvisor on reputation within the hotel industry. TripAdvisor encapsulates key themes in establishing an online reputation strategy in an evolving digital landscape. Design/methodology/approach Through the use of an exploratory case study, data were gathered primarily by means of a series of expert interviews within the hotel industry in Cyprus, today a mature holiday destination in Europe. Further data collection included a document search of presentations, annual reports, past surveys and sales and marketing literature from the examined industry. Findings Hotel communication practitioners are fully aware of the impact of social media in managing reputation. Constant monitoring, prompt responses, training and transparency were identified as key factors. Online reputation management needs to be taken into consideration when designing a comprehensive integrated communication strategy. Research limitations/implications Congruence amongst interviewees in certain areas could be on account of the homogeneity of practitioners, of their background and training and of similar organisational cultures across the locale of study. This leads to limits in the generalisations from this study’s findings. Practical implications Encouragement and training of employees were amongst the primary suggestions that emerged. An internal and external environmental scan, recognising possible strengths, weaknesses, opportunities and threats, which could assist in the effective engagement and monitoring of the organisation’s online presence, were also suggested. Originality/value The uniqueness of the study lies in its exploration of reputation management of a well-known traveller’s platform by addressing social media content in both a proactive and reactive manner.

Download Full-text

StyleGANs and Transfer Learning for Generating Synthetic Images in Industrial Applications

Symmetry ◽

10.3390/sym13081497 ◽

2021 ◽

Vol 13 (8) ◽

pp. 1497

Author(s):

Harold Achicanoy ◽

Deisy Chaves ◽

Maria Trujillo

Keyword(s):

Deep Learning ◽

Transfer Learning ◽

Data Augmentation ◽

Industrial Applications ◽

Generative Models ◽

Training Data ◽

Generative Adversarial Networks ◽

Augmentation Strategy ◽

Synthetic Images ◽

The Impact

Deep learning applications on computer vision involve the use of large-volume and representative data to obtain state-of-the-art results due to the massive number of parameters to optimise in deep models. However, data are limited with asymmetric distributions in industrial applications due to rare cases, legal restrictions, and high image-acquisition costs. Data augmentation based on deep learning generative adversarial networks, such as StyleGAN, has arisen as a way to create training data with symmetric distributions that may improve the generalisation capability of built models. StyleGAN generates highly realistic images in a variety of domains as a data augmentation strategy but requires a large amount of data to build image generators. Thus, transfer learning in conjunction with generative models are used to build models with small datasets. However, there are no reports on the impact of pre-trained generative models, using transfer learning. In this paper, we evaluate a StyleGAN generative model with transfer learning on different application domains—training with paintings, portraits, Pokémon, bedrooms, and cats—to generate target images with different levels of content variability: bean seeds (low variability), faces of subjects between 5 and 19 years old (medium variability), and charcoal (high variability). We used the first version of StyleGAN due to the large number of publicly available pre-trained models. The Fréchet Inception Distance was used for evaluating the quality of synthetic images. We found that StyleGAN with transfer learning produced good quality images, being an alternative for generating realistic synthetic images in the evaluated domains.

Download Full-text