Transfer Learning-based Approach Research Articles

3D_DEN: Open-Ended 3-D Object Recognition Using Dynamically Expandable Networks

Service robots, in general, have to work independently and adapt to the dynamic changes in the environment. One important aspect in such scenarios is to continually learn to recognize newer object categories when they become available. This combines two main research problems namely continual learning and 3D object recognition. Most of the existing research approaches include the use of deep Convolutional Neural Networks (CNNs) focusing on image datasets. A modified approach might be needed for continually learning 3D object categories. A major concern in using CNNs is the problem of catastrophic forgetting when a model tries to learn a new task. Despite various recent proposed solutions to mitigate this problem, there still exist a few side-effects (such as time/computational complexity) of such solutions. We propose a model capable of learning 3D object categories in an open-ended fashion by employing a deep transfer learning-based approach combined with dynamically expandable layers, which also makes sure that these side-effects are minimized to a great extent. We show that this model sets a new state-of-the-art standard with regards to accuracy and also substantially minimizes computational complexity.

A new transfer learning based approach to magnification dependent and independent classification of breast cancer in histopathological images

The visual analysis of histopathological images is the gold standard for diagnosing breast cancer, yet a strenuous and an intricate task that requires years of pathologist training. Therefore, automating this task using computer-aided diagnosis (CAD) is highly expected. This paper proposes a new transfer learning-based approach to automated classification of breast cancer from histopathological images, including magnification dependent (MD) and magnification independent (MI) binary and eight-class classifications. We apply the deep neural network ResNet-18 to this problem, which is pre-trained on ImageNet, a large dataset of common images. We then design our transfer learning method to refine the network on histopathological images. Our transfer learning method is based on block-wise fine-tuning strategy; in which we make the last two residual blocks of the deep network model more domain-specific to our target data. It substantially helps to avoid over-fitting and speed up the training. Furthermore, we strengthen the adaptability of the proposed approach by using global contrast normalization (GCN) based on the target’s data values and three-fold data augmentation on training data. The experimental results of MD and MI binary and eight-class classifications on the publicly available BreaKHis dataset demonstrate that our approach is promising and effective, outperforming recent state-of-the-art MD and MI counterparts by a fair margin.

Automated Deep Transfer Learning-Based Approach for Detection of COVID-19 Infection in Chest X-rays

The most widely used novel coronavirus (COVID-19) detection technique is a real-time polymerase chain reaction (RT-PCR). However, RT-PCR kits are costly and take 6-9 hours to confirm infection in the patient. Due to less sensitivity of RT-PCR, it provides high false-negative results. To resolve this problem, radiological imaging techniques such as chest X-rays and computed tomography (CT) are used to detect and diagnose COVID-19. In this paper, chest X-rays is preferred over CT scan. The reason behind this is that X-rays machines are available in most of the hospitals. X-rays machines are cheaper than the CT scan machine. Besides this, X-rays has low ionizing radiations than CT scan. COVID-19 reveals some radiological signatures that can be easily detected through chest X-rays. For this, radiologists are required to analyze these signatures. However, it is a time-consuming and error-prone task. Hence, there is a need to automate the analysis of chest X-rays. The automatic analysis of chest X-rays can be done through deep learning-based approaches, which may accelerate the analysis time. These approaches can train the weights of networks on large datasets as well as fine-tuning the weights of pre-trained networks on small datasets. However, these approaches applied to chest X-rays are very limited. Hence, the main objective of this paper is to develop an automated deep transfer learning-based approach for detection of COVID-19 infection in chest X-rays by using the extreme version of the Inception (Xception) model. Extensive comparative analyses show that the proposed model performs significantly better as compared to the existing models.

A new transfer learning-based method for label proportions problem

Learning with label proportions (LLP), which only provides the unlabeled instances in the bag and the bag’s label proportion, has been widely studied recently. However, most of the existing LLP methods do not consider the knowledge transfer from the source task to the target task. In addition, in the process of the collecting data, the data may be corrupted by noise, and this always leads to the uncertain data in its representation. This paper proposes a new transfer learning-based approach for the problem of learning with label proportions, which is called TL-LLP in brief, to transfer knowledge from the source task to the target task where both the source and target tasks contain uncertain data. We first formulate the objective model to deal with transfer learning and uncertain data for the label proportions problem at the same time. We then propose an iterative framework to solve the proposed objective model and obtain the accurate classifier for the target task. Extensive experiments have shown that the proposed TL-LLP method can obtain better performance and is less sensitive to noise compared with the existing LLP methods.

Deep Transfer Learning for Source Code Modeling

In recent years, deep learning models have shown great potential in source code modeling and analysis. Generally, deep learning-based approaches are problem-specific and data-hungry. A challenging issue of these approaches is that they require training from scratch for a different related problem. In this work, we propose a transfer learning-based approach that significantly improves the performance of deep learning-based source code models. In contrast to traditional learning paradigms, transfer learning can transfer the knowledge learned in solving one problem into another related problem. First, we present two recurrent neural network-based models RNN and GRU for the purpose of transfer learning in the domain of source code modeling. Next, via transfer learning, these pre-trained (RNN and GRU) models are used as feature extractors. Then, these extracted features are combined into attention learner for different downstream tasks. The attention learner leverages from the learned knowledge of pre-trained models and fine-tunes them for a specific downstream task. We evaluate the performance of the proposed approach with extensive experiments with the source code suggestion task. The results indicate that the proposed approach outperforms the state-of-the-art models in terms of accuracy, precision, recall and F-measure without training the models from scratch.

A Spectrogram-Based Deep Feature Assisted Computer-Aided Diagnostic System for Parkinson’s Disease

Parkinson’s disease is a neural degenerative disease. It slowly progresses from mild to severe stage, resulting in the degeneration of dopamine cells of neurons. Due to the deficiency of dopamine cells in the brain, it leads to a motor (tremor, slowness, impaired posture) and non-motor (speech, olfactory) defects in the body. Early detection of Parkinson’s disease is a difficult chore as the symptoms of disease appear overtime. However, different diagnostic systems have contributed towards disease detection by considering gait, tremor and speech characteristics. Recent work has shown that speech impairments can be considered as a possible predictor for Parkinson’s disease classification and remains an open research area. The speech signals show major differences and variations for Parkinson patients as compared to normal human beings. Therefore, variation in speech should be modeled using acoustic features to identify these variations. In this research, we propose three methods- the first method employs a transfer learning-based approach using spectrograms of speech recordings, the second method evaluates deep features extracted from speech spectrograms using machine learning classifiers and the third method evaluates simple acoustic feature of recordings using machine learning classifiers. The proposed frameworks are evaluated on a Spanish dataset pc-Gita. The results show that the second framework shows promising results with deep features. The highest 99.7% accuracy on vowel $\backslash \text{o}\backslash $ and read text is observed using a multilayer perceptron. Whereas 99.1% accuracy observed on vowel $\backslash \text{i}\backslash $ deep features using random forest. The deep feature-based method performs better as compared to simple acoustic features and transfer learning approaches. The proposed methodology outperforms the existing techniques on the pc-Gita dataset for Parkinson’s disease detection.

Simple and Easy: Transfer Learning-Based Attacks to Text CAPTCHA

CAPTCHA, or Completely Automated Public Turing Tests to Tell Computers and Humans Apart, is a common mechanism used to protect commercial accounts from malicious computer bots, and the most widely used scheme is text-based CAPTCHA. In recent years, newly emerged deep learning techniques have achieved high accuracy and speed in attacking text-based CAPTCHAs. However, most of the existing attacks have various disadvantages, the attack process made high complexity or manually collecting and labeling a large number of samples to train a deep learning recognition model is time-consuming and expensive. In this paper, we propose a transfer learning-based approach that greatly reduces the attack complexity and the cost of labeling samples, specifically, by pre-training the model with randomly generated samples and fine-tuning the pre-trained model with a small number of real-world samples. To evaluate our attack, we tested 25 online CAPTCHAs achieving success rates ranging from 36.3% to 96.9%. To further explore the effect of the training sample characteristics on the attack accuracy, we elaborately imitate some samples and apply a generative adversarial network to refine the samples, sequentially we use these two kinds of generated samples to pre-train the models, respectively. The experimental results demonstrate that the similarity between randomly generated samples and elaborately imitated samples has a negligible impact on the attack accuracy. Instead, transfer learning is the key factor; it reduces the cost of data preparation while preserving the model's attack accuracy.

Super resolution land cover mapping of hyperspectral images using the deep image prior-based approach

ABSTRACTDue to the instantaneous field-of-view (IFOV) of the sensor and diversity of land cover types, some pixels, usually named mixed pixels, contain more than one land cover type. Soft classification can predict the portion of each land cover type in mixed pixels in the absence of spatial distribution. The spatial distribution information in mixed pixels can be solved by super resolution mapping (SRM). Typically, SRM involves two steps: soft class value estimation, which is similar to the image super resolution of image restoration, and land cover allocation. A new SRM approach utilizes a deep image prior (DIP) strategy combined with a super resolution convolutional neural network (SRCNN) to estimate fine resolution fraction images for each land cover type; then, a simple and efficient classifier is used to allocate subpixel land cover types under the constraint of the generated fine fraction images. The proposed approach can use prior information of input images to update network parameters and no longer require training data. Experiments on three different cases demonstrate that the subpixel classification accuracy of the proposed DIP-based SRM approach is significantly better than the three conventional SRM approaches and a transfer learning-based neural network SRM approach. In addition, the DIP-SRM approach performs very robustly about small-area objects within multiple land cover types and significantly reduces soft classification uncertainty. The results of this paper provide an extension for utilizing SRCNN to address SRM issues in hyperspectral images.

A Two-Stage Transfer Learning-Based Deep Learning Approach for Production Progress Prediction in IoT-Enabled Manufacturing

In make-to-order manufacturing enterprises, accurate production progress (PP) prediction is an important basis for dynamic production process optimization and on-time delivery of orders. The implementation of Internet of Things (IoT) makes it possible to take real-time production state as an important factor affecting PP. In the IoT-enabled workshop, a two-stage transfer learning-based prediction method using both historical production data and real-time state data is proposed to solve the problem of low-prediction accuracy and poor generalization performance caused by insufficient data of target order. The deep autoencoder (DAE) model with transfer learning is designed to extract the generalized features of target order in the first stage, which uses bootstrap sampling to avoid over fitting. The deep belief network (DBN) model with transfer learning is constructed to fit the nonlinear relation for PP prediction in the second stage. A real case from an IoT enabled machining workshop is taken to validate the performance of the proposed method over the other methods such as DBN, deep neural network.

Surface EMG signals and deep transfer learning-based physical action classification

Human physical action classification is an emerging area of research for human-to-machine interaction, which can help to disable people to interact with real world, and robotics application. EMG signals measure the electrical activity muscular systems, which involved in physical action of human. EMG signals provide more information related to physical action. In this paper, we proposed deep transfer learning-based approach of human action classification using surface EMG signals. The surface EMG signals are represented by time–frequency image (TFI) by using short-time Fourier transform. TFI is used as input to pre-trained convolutional neural network models, namely AlexNet and VGG16, for deep feature extraction, and support vector machine (SVM) classifier is used for classification of physical action of EMG signals. Also, the fine-tuning of the pre-trained AlexNet model is also considered. The experimental results show that deep feature extraction and SVM classification method and fine-tuning have obviously improved the classification accuracy when compared with various results from the literature. The 99.04% accuracy score is obtained with AlexNet fc6 + AlexNet fc7 + VGG16 fc6 + VGG16 fc7 deep feature concatenation and SVM classification. 98.65% accuracy score is performed by fine-tuning of the AlexNet model. We also compare the obtained results with some of the existing methods. The comparisons show that the deep feature concatenation and SVM classification method provide better classification accuracy than the compared methods.

Relative Afferent Pupillary Defect Screening Through Transfer Learning.

Abnormalities in pupillary light reflex can indicate optic nerve disorders that may lead to permanent visual loss if not diagnosed in an early stage. In this study, we focus on relative afferent pupillary defect (RAPD), which is based on the difference between the reactions of the eyes when they are exposed to light stimuli. Incumbent RAPD assessment methods are based on subjective practices that can lead to unreliable measurements. To eliminate subjectivity and obtain reliable measurements, we introduced an automated framework to detect RAPD. For validation, we conducted a clinical study with lab-on-a-headset, which can perform automated light reflex test. In addition to benchmarking handcrafted algorithms, we proposed a transfer learning-based approach that transformed a deep learning-based generic object recognition algorithm into a pupil detector. Based on the conducted experiments, proposed algorithm RAPDNet can achieve a sensitivity and a specificity of 90.6% over 64 test cases in a balanced set, which corresponds to an AUC of 0.929 in ROC analysis. According to our benchmark with three handcrafted algorithms and nine performance metrics, RAPDNet outperforms all other algorithms in every performance category.

Automatic detection of passable roads after floods in remote sensed and social media data

This paper addresses the problem of floods classification and floods aftermath detection based on both social media and satellite imagery. Automatic detection of disasters such as floods is still a very challenging task. The focus lies on identifying passable routes or roads during floods. Two novel solutions are presented, which were developed for two corresponding tasks at the MediaEval 2018 benchmarking challenge. The tasks are (i) identification of images providing evidence for road passability and (ii) differentiation and detection of passable and non-passable roads in images from two complementary sources of information. For the first challenge, we mainly rely on object and scene-level features extracted through multiple deep models pre-trained on the ImageNet and Places datasets. The object and scene-level features are then combined using early, late and double fusion techniques. To identify whether or not it is possible for a vehicle to pass a road in satellite images, we rely on Convolutional Neural Networks and a transfer learning-based classification approach. The evaluation of the proposed methods is carried out on the large-scale datasets provided for the benchmark competition. The results demonstrate significant improvement in the performance over the recent state-of-art approaches.

A Transfer Learning Approach for Microstructure Reconstruction and Structure-property Predictions

Stochastic microstructure reconstruction has become an indispensable part of computational materials science, but ongoing developments are specific to particular material systems. In this paper, we address this generality problem by presenting a transfer learning-based approach for microstructure reconstruction and structure-property predictions that is applicable to a wide range of material systems. The proposed approach incorporates an encoder-decoder process and feature-matching optimization using a deep convolutional network. For microstructure reconstruction, model pruning is implemented in order to study the correlation between the microstructural features and hierarchical layers within the deep convolutional network. Knowledge obtained in model pruning is then leveraged in the development of a structure-property predictive model to determine the network architecture and initialization conditions. The generality of the approach is demonstrated numerically for a wide range of material microstructures with geometrical characteristics of varying complexity. Unlike previous approaches that only apply to specific material systems or require a significant amount of prior knowledge in model selection and hyper-parameter tuning, the present approach provides an off-the-shelf solution to handle complex microstructures, and has the potential of expediting the discovery of new materials.

Understanding Social Networks Using Transfer Learning

A detailed understanding of users contributes to the understanding of the Web's evolution, and to the development of Web applications. Although for new Web platforms such a study is especially important, it is often jeopardized by the lack of knowledge about novel phenomena due to the sparsity of data. Akin to human transfer of experiences from one domain to the next, transfer learning as a subfield of machine learning adapts knowledge acquired in one domain to a new domain. We systematically investigate how the concept of transfer learning may be applied to the study of users on newly created (emerging) Web platforms, and propose our transfer learning-based approach, TraNet. We show two use cases where TraNet is applied to tasks involving the identification of user trust and roles on different Web platforms. We compare the performance of TraNet with other approaches and find that our approach can best transfer knowledge on users across platforms in the given tasks.

A Learning-Based Approach to Caching in Heterogenous Small Cell Networks

A heterogenous network with base stations (BSs), small base stations (SBSs) and users distributed according to independent Poisson point processes is considered. SBS nodes are assumed to possess high storage capacity and to form a distributed caching network. Popular files are stored in local caches of SBSs, so that a user can download the desired files from one of the SBSs in its vicinity. The offloading-loss is captured via a cost function that depends on the random caching strategy proposed here. The popularity profile of cached content is unknown and estimated using instantaneous demands from users within a specified time interval. An estimate of the cost function is obtained from which an optimal random caching strategy is devised. The training time to achieve an $\epsilon>0$ difference between the achieved and optimal costs is finite provided the user density is greater than a predefined threshold, and scales as $N^2$, where $N$ is the support of the popularity profile. A transfer learning-based approach to improve this estimate is proposed. The training time is reduced when the popularity profile is modeled using a parametric family of distributions; the delay is independent of $N$ and scales linearly with the dimension of the distribution parameter.