Semi-supervised Deep Learning Approach Research Articles

Ensemble-based approach for semisupervised learning in remote sensing

Semisupervised learning (SSL) techniques explore the progressive discovery of the hidden latent data structure by propagating supervised information on unlabeled data, which are thereafter used to reinforce learning. These schemes are beneficial in remote sensing, where thousands of new images are added every day, and manual labeling results are prohibitive. Our work introduces an ensemble-based semisupervised deep learning approach that initially takes a subset of labeled data Dl, which represents the latent structure of the data and progressively propagates labels automatically from an expanding set of unlabeled data Du. The ensemble is a set of classifiers whose predictions are collated to derive a consolidated prediction. Only those data having a high-confidence prediction are considered as newly generated labels. The proposed approach was exhaustively validated on four public datasets, achieving appreciable results compared to the state-of-the-art methods in most of the evaluated configurations. For all datasets, the proposed approach achieved a classification F1-score and recall of up to 90%, on average. The SSL and recursive scheme also demonstrated an average gain of ∼2 % at the last training stage in such large datasets.

Semi-Supervised Spatiotemporal Deep Learning for Intrusions Detection in IoT Networks

The rapid growth of the Internet of Things (IoT) technologies has generated a huge amount of traffic that can be exploited for detecting intrusions through IoT networks. Despite the great effort made in annotating IoT traffic records, the number of labeled records is still very small, increasing the difficulty in recognizing attacks and intrusions. This study introduces a semi-supervised deep learning approach for intrusion detection (SS-Deep-ID), in which we propose a multiscale residual temporal convolutional (MS-Res) module to finetune the network capability in learning spatiotemporal representations. An improved traffic attention (TA) mechanism is introduced to estimate the importance score that helps the model to concentrate on important information during learning. Furthermore, a hierarchical semi-supervised training method is introduced which takes into account the sequential characteristics of the IoT traffic data during training. The proposed SS-Deep-ID is easily integrated into a fog-enabled IoT network to offer efficient real-time intrusion detection. Finally, empirical evaluations on two recent data sets (CIC-IDS2017 and CIC-IDS2018) demonstrate that SS-Deep-ID improves the efficiency of intrusion detection and increases the robustness of performance while maintaining computational efficiency.

Building Extraction from Very-High-Resolution Remote Sensing Images Using Semi-Supervised Semantic Edge Detection

The automated detection of buildings in remote sensing images enables understanding the distribution information of buildings, which is indispensable for many geographic and social applications, such as urban planning, change monitoring and population estimation. The performance of deep learning in images often depends on a large number of manually labeled samples, the production of which is time-consuming and expensive. Thus, this study focuses on reducing the number of labeled samples used and proposing a semi-supervised deep learning approach based on an edge detection network (SDLED), which is the first to introduce semi-supervised learning to the edge detection neural network for extracting building roof boundaries from high-resolution remote sensing images. This approach uses a small number of labeled samples and abundant unlabeled images for joint training. An expert-level semantic edge segmentation model is trained based on labeled samples, which guides unlabeled images to generate pseudo-labels automatically. The inaccurate label sets and manually labeled samples are used to update the semantic edge model together. Particularly, we modified the semantic segmentation network D-LinkNet to obtain high-quality pseudo-labels. Specifically, the main network architecture of D-LinkNet is retained while the multi-scale fusion is added in its second half to improve its performance on edge detection. The SDLED was tested on high-spatial-resolution remote sensing images taken from Google Earth. Results show that the SDLED performs better than the fully supervised method. Moreover, when the trained models were used to predict buildings in the neighboring counties, our approach was superior to the supervised way, with line IoU improvement of at least 6.47% and F1 score improvement of at least 7.49%.

A semi-supervised deep learning approach for predicting the functional effects of genomic non-coding variations

BackgroundUnderstanding the functional effects of non-coding variants is important as they are often associated with gene-expression alteration and disease development. Over the past few years, many computational tools have been developed to predict their functional impact. However, the intrinsic difficulty in dealing with the scarcity of data leads to the necessity to further improve the algorithms. In this work, we propose a novel method, employing a semi-supervised deep-learning model with pseudo labels, which takes advantage of learning from both experimentally annotated and unannotated data.ResultsWe prepared known functional non-coding variants with histone marks, DNA accessibility, and sequence context in GM12878, HepG2, and K562 cell lines. Applying our method to the dataset demonstrated its outstanding performance, compared with that of existing tools. Our results also indicated that the semi-supervised model with pseudo labels achieves higher predictive performance than the supervised model without pseudo labels. Interestingly, a model trained with the data in a certain cell line is unlikely to succeed in other cell lines, which implies the cell-type-specific nature of the non-coding variants. Remarkably, we found that DNA accessibility significantly contributes to the functional consequence of variants, which suggests the importance of open chromatin conformation prior to establishing the interaction of non-coding variants with gene regulation.ConclusionsThe semi-supervised deep learning model coupled with pseudo labeling has advantages in studying with limited datasets, which is not unusual in biology. Our study provides an effective approach in finding non-coding mutations potentially associated with various biological phenomena, including human diseases.

Semi-supervised deep learning approach to break common CAPTCHAs

Manual data annotation is a time consuming activity. A novel strategy for automatic training of the CAPTCHA breaking system with no manual dataset creation is presented in this paper. We demonstrate the feasibility of the attack against a text-based CAPTCHA scheme utilizing similar network infrastructure used for Denial of Service attacks. The main goal of our research is to present a possible vulnerability in CAPTCHA systems when combining the brute-force attack with transfer learning. The classification step utilizes a simple convolutional neural network with 15 layers. Training stage uses automatically prepared dataset created without any human intervention and transfer learning for fine-tuning the deep neural network classifier. The designed system for breaking text-based CAPTCHAs achieved 80% classification accuracy after 6 fine-tuning steps for a 5 digit text-based CAPTCHA system. The results presented in this paper suggest, that even the simple attack with a large number of attacking computers can be an effective alternative to current CAPTCHA breaking systems.

Detecting False Data Injection Attacks in Smart Grids: A Semi-Supervised Deep Learning Approach

The dependence on advanced information and communication technology increases the vulnerability in smart grids under cyber-attacks. Recent research on unobservable false data injection attacks (FDIAs) reveals the high risk of secure system operation, since these attacks can bypass current bad data detection mechanisms. To mitigate this risk, this paper proposes a data-driven learning-based algorithm for detecting unobservable FDIAs in distribution systems. We use autoencoders for efficient dimension reduction and feature extraction of measurement datasets. Further, we integrate the autoencoders into an advanced generative adversarial network (GAN) framework, which successfully detects anomalies under FDIAs by capturing the unconformity between abnormal and secure measurements. Also, considering that the datasets collected from practical power systems are partially labeled due to expensive labeling costs and missing labels, the proposed method only requires a few labeled measurement data in addition to unlabeled data for training. Numerical simulations in three-phase unbalanced IEEE 13-bus and 123-bus distribution systems validate the detection accuracy and efficiency of this method.

Solo: Doublet Identification in Single-Cell RNA-Seq via Semi-Supervised Deep Learning.

Single-cell RNA sequencing (scRNA-seq) measurements of gene expression enable an unprecedented high-resolution view into cellular state. However, current methods often result in two or more cells that share the same cell-identifying barcode; these "doublets" violate the fundamental premise of single-cell technology and can lead to incorrect inferences. Here, we describe Solo, a semi-supervised deep learning approach that identifies doublets with greater accuracy than existing methods. Solo embeds cells unsupervised using a variational autoencoder and then appends a feed-forward neural network layer to the encoder to form a supervised classifier. We train this classifier to distinguish simulated doublets from the observed data. Solo can be applied in combination with experimental doublet detection methods to further purify scRNA-seq data to true single cells. It is freely available from https://github.com/calico/solo. A record of this paper's transparent peer review process is included in the Supplemental Information.

Semi-Supervised Deep Learning Approach for Transportation Mode Identification Using GPS Trajectory Data

Identification of travelers' transportation modes is a fundamental step for various problems that arise in the domain of transportation such as travel demand analysis, transport planning, and traffic management. In this paper, we aim to identify travelers' transportation modes purely based on their GPS trajectories. First, a segmentation process is developed to partition a user's trip into GPS segments with only one transportation mode. A majority of studies have proposed mode inference models based on hand-crafted features, which might be vulnerable to traffic and environmental conditions. Furthermore, the classification task in almost all models have been performed in a supervised fashion while a large amount of unlabeled GPS trajectories has remained unused. Accordingly, we propose a deep SEmi-Supervised Convolutional Autoencoder (SECA) architecture that can not only automatically extract relevant features from GPS segments but also exploit useful information in unlabeled data. The SECA integrates a convolutional-deconvolutional autoencoder and a convolutional neural network into a unified framework to concurrently perform supervised and unsupervised learning. The two components are simultaneously trained using both labeled and unlabeled GPS segments, which have already been converted into an efficient representation for the convolutional operation. An optimum schedule for varying the balancing parameters between reconstruction and classification errors are also implemented. The performance of the proposed SECA model, trip segmentation, the method for converting a raw trajectory into a new representation, the hyperparameter schedule, and the model configuration are evaluated by comparing to several baselines and alternatives for various amounts of labeled and unlabeled data. Our experimental results demonstrate the superiority of the proposed model over the state-of-the-art semi-supervised and supervised methods with respect to metrics such as accuracy and F-measure.

A semi-supervised deep learning approach for circular hole detection on composite parts

This paper introduces the usage of semi-supervised learning to obtain competitive detection accuracy of measuring drilled holes on composite parts with very limited noisy training data. An improved texture segmentation algorithm based on local binary patterns algorithm is proposed, named local exponential patterns. The algorithm divides the image texture into nine levels, of which the highest level of texture is selected for contour extraction. An ellipse fitting method is used to fit the target contours and vote for the candidate ellipses. The regions inside the candidate ellipses are taken as the semi-supervised semantic label for images. A new loss named round loss is proposed, and a superior circle segmentation model was trained by learning from incompletely annotated data. To verify the effectiveness of the method, experiments were conducted with the drilled holes on the composite parts. The results show that the proposed semi-supervised deep learning approach is exceedingly suitable for circle detection of holes with different texture information commonly found in robotic drilling. Massive data labeling can be completely avoided with proposed method. The measurement accuracy can reach 0.03 mm, which can meet the visual measurement requirements of the circular holes on composite parts in the robotic drilling system.

CT Super-Resolution GAN Constrained by the Identical, Residual, and Cycle Learning Ensemble (GAN-CIRCLE).

In this paper, we present a semi-supervised deep learning approach to accurately recover high-resolution (HR) CT images from low-resolution (LR) counterparts. Specifically, with the generative adversarial network (GAN) as the building block, we enforce the cycle-consistency in terms of the Wasserstein distance to establish a nonlinear end-to-end mapping from noisy LR input images to denoised and deblurred HR outputs. We also include the joint constraints in the loss function to facilitate structural preservation. In this process, we incorporate deep convolutional neural network (CNN), residual learning, and network in network techniques for feature extraction and restoration. In contrast to the current trend of increasing network depth and complexity to boost the imaging performance, we apply a parallel 1×1 CNN to compress the output of the hidden layer and optimize the number of layers and the number of filters for each convolutional layer. The quantitative and qualitative evaluative results demonstrate that our proposed model is accurate, efficient and robust for super-resolution (SR) image restoration from noisy LR input images. In particular, we validate our composite SR networks on three large-scale CT datasets, and obtain promising results as compared to the other state-of-the-art methods.

Individual Tree-Crown Detection in RGB Imagery Using Semi-Supervised Deep Learning Neural Networks

Remote sensing can transform the speed, scale, and cost of biodiversity and forestry surveys. Data acquisition currently outpaces the ability to identify individual organisms in high resolution imagery. We outline an approach for identifying tree-crowns in RGB imagery while using a semi-supervised deep learning detection network. Individual crown delineation has been a long-standing challenge in remote sensing and available algorithms produce mixed results. We show that deep learning models can leverage existing Light Detection and Ranging (LIDAR)-based unsupervised delineation to generate trees that are used for training an initial RGB crown detection model. Despite limitations in the original unsupervised detection approach, this noisy training data may contain information from which the neural network can learn initial tree features. We then refine the initial model using a small number of higher-quality hand-annotated RGB images. We validate our proposed approach while using an open-canopy site in the National Ecological Observation Network. Our results show that a model using 434,551 self-generated trees with the addition of 2848 hand-annotated trees yields accurate predictions in natural landscapes. Using an intersection-over-union threshold of 0.5, the full model had an average tree crown recall of 0.69, with a precision of 0.61 for the visually-annotated data. The model had an average tree detection rate of 0.82 for the field collected stems. The addition of a small number of hand-annotated trees improved the performance over the initial self-supervised model. This semi-supervised deep learning approach demonstrates that remote sensing can overcome a lack of labeled training data by generating noisy data for initial training using unsupervised methods and retraining the resulting models with high quality labeled data.

VUNet: Dynamic Scene View Synthesis for Traversability Estimation Using an RGB Camera

We present VUNet, a novel view(VU) synthesis method for mobile robots in dynamic environments, and its application to the estimation of future traversability. Our method predicts future images for given virtual robot velocity commands using only RGB images at previous and current time steps. The future images result from applying two types of image changes to the previous and current images: first, changes caused by different camera pose. Second, changes due to the motion of the dynamic obstacles. We learn to predict these two types of changes disjointly using two novel network architectures, SNet and DNet. We combine SNet and DNet to synthesize future images that we pass to our previously presented method GONet [N. Hirose, A. Sadeghian, M. Vazquez, P. Goebel, and S. Savarese, “Gonet: A semi-supervised deep learning approach for traversability estimation,” in Proc. IEEE International Conference on Intelligent Robots and Systems, 2018, pp. 3044–3051] to estimate the traversable areas around the robot. Our quantitative and qualitative evaluation indicate that our approach for view synthesis predicts accurate future images in both static and dynamic environments. We also show that these virtual images can be used to estimate future traversability correctly. We apply our view synthesis-based traversability estimation method to two applications for assisted teleoperation.

Utilizing Unlabeled Data to Detect Electricity Fraud in AMI: A Semisupervised Deep Learning Approach.

As nontechnical losses in power systems have recently become a global concern, electricity fraud detection models attracted increasing academic interest. The wide application of smart meters has offered more possibility to detecting fraud from user's consumption patterns. However, the performances of existing consumption-based electricity fraud detection models are still not satisfactory enough for practice, partly due to their limited ability to handle high-dimensional data. In this paper, a deep-learning-based model is developed for detecting electricity fraud in the advanced metering infrastructure, namely, the multitask feature extracting fraud detector (MFEFD). The deep architecture has brought MFEFD a powerful ability to handle high-dimensional input, through which consumption patterns inside load profiles can be effectively extracted. Another challenge is that the insufficiency of labeled data has restricted the generalization of existing models since they are mostly based on supervised learning and labeled data. MFEFD is trained in a semisupervised manner, in which multitask training was implemented to combine the supervised and unsupervised training, so that both the knowledge from unlabeled and labeled data can be made use of. Real-world-data-based case studies have demonstrated MFEFD's high detection performance, robustness, privacy preservation, and practicability.

An EM-based semi-supervised deep learning approach for semantic segmentation of histopathological images from radical prostatectomies

Automated Gleason grading is an important preliminary step for quantitative histopathological feature extraction. Different from the traditional task of classifying small pre-selected homogeneous regions, semantic segmentation provides pixel-wise Gleason predictions across an entire slide. Deep learning-based segmentation models can automatically learn visual semantics from data, which alleviates the need for feature engineering. However, performance of deep learning models is limited by the scarcity of large-scale fully annotated datasets, which can be both expensive and time-consuming to create. One way to address this problem is to leverage external weakly labeled datasets to augment models trained on the limited data. In this paper, we developed an expectation maximization-based approach constrained by an approximated prior distribution in order to extract useful representations from a large number of weakly labeled images generated from low-magnification annotations. This method was utilized to improve the performance of a model trained on a limited fully annotated dataset. Our semi-supervised approach trained with 135 fully annotated and 1800 weakly annotated tiles achieved a mean Jaccard Index of 49.5% on an independent test set, which was 14% higher than the initial model trained only on the fully annotated dataset.