Search

Coronavirus disease (COVID-19) is an infectious respiratory illness caused by Severe Acute Respiratory Syndrome CoronaVirus 2 (SARS-CoV-2). The virus is new and has caused an unprecedented pandemic that has led into unforeseeable circumstances. Due to lack of knowledge, medical fraternity is facing a lot of hurdles in the detection and treatment of this disease which is caused by coronavirus. It is immensely crucial to handle the prevailing situation for the sake of healthier lives in future. In present scenario, data recorded from the occurrences of the disease worldwide can play a vital role in creating a better understanding of its nature. Based on data, artificial intelligence techniques like machine learning provide useful insights and solutions for real world problems that are otherwise incomprehensible for human. Specifically, clustering is an unsupervised machine learning technique that uses unlabelled data to identify different behaviour, groups and categories of observations. This paper presents a machine learning based clustering approach to identify different clusters of symptomatic profiles of patients suffering from covid-19. Dividing the set of medical symptoms experienced by covid-19 patients into distinct clusters will facilitate an understanding of the different kinds of effects this disease has on human beings. It will also enable development of suitable and effective treatments.

Whilst convolutional neural networks (CNN) offer state-of-the-art performance for classification and detection tasks in computer vision, their successful adoption in defence applications is limited by the cost of labelled data and the inability to use crowd sourcing due to classification issues. Popular approaches to solve this problem use the expansive labelled data for training. It would be more cost-efficient to learn representations from the unlabelled data whilst leveraging labelled data from existing datasets, as empirically the performance of supervised learning is far greater than unsupervised-learning. In this paper we investigate the benefits of mixing Domain Adaptation and semi-supervised learning to train CNNs and showcase using adversarial training to tackle this issue.

Automatic surface inspection (ASI) to identify defects in manufactured items plays an important role in ensuring the production quality in industrial manufacturing processes. Various approaches have been proposed for this purpose, and the majority of them use supervised learning. Supervised learning requires labelled data for training. Obtaining a large amount of labelled data is a difficult, and time consuming task. On the other hand, Semi-supervised learning approaches become popular for ASI, as they make use of both labelled and unlabelled data. In this work, we propose a Convolutional Neural Network based semi-supervised learning approach for the recognition of steel surface defects. Our approach predicts the labels of the unlabelled data, and weights them based on their prediction confidence. These weighted samples are then used with their corresponding predicted labels, together with the labelled data for training the network. Our approach mainly differs from the existing approaches in the way the unlabelled samples are weighted when training the network. We propose to weight the samples based on how confidently they are predicted. We propose a margin-based approach to determine the prediction confidence. Experimental results on a public steel surface detection dataset (NEU surface defects) show that the proposed method can achieve a state-of-the-art accuracy of 99.15 ±0.08%, which is competitive compared to the performance achieved by fully supervised deep learning approaches, but ours with only 10% of labelled training data compared to the supervised learning approaches. In addition, comparison with recently proposed semi supervised learning approaches for ASI shows the effectiveness of our approach.

Objective: Laryngopharyngeal Reflux Disease (LPRD) is prevalent and has a range of symptoms. However, diagnosing LPRD is difficult because of the lack of specific symptoms or clinical gold standard. An objective and reliable test, which does not rely on a perfect clinical gold standard, is required in clinical practice. Methods: 60 normal volunteers and 74 confirmed Laryngopharyngeal Reflux (LPR) patients were labelled based on the combined consideration of the reflux symptom index, reflux finding score, 24h oropharyngeal pH monitoring and results of anti-reflux treatment. 72 candidate features were extracted from pH recordings and the most efficient feature combination was detected using a stepwise wrapper method. The labelled data were combined with 1552 unlabelled data for feature selection and model training using semi-supervised learning. A latent class model method was used to assess the proposed model based on 64 additional validation data and an imperfect clinical reference test. Results: A new score (named W score), which significantly improved the sensitivity of the LPRD test (82.67% vs. 24.09%) and has a relatively high specificity (80.19%), was proposed. W score concurs with the complicated clinical test. Conclusion: W score which significantly improves LPRD diagnostic efficiency, can aid the clinical diagnosis of LPRD. W score provides an objective, efficient, and reliable indicator for the application of anti-reflux treatment in clinical practice. Significance: More LPRD patients can benefit from anti-reflux treatment in clinical practice and fewer patients may suffer from, for example, the side effects of unnecessary long-term Proton Pump Inhibitors (PPI) treatments.

In this work, we study music/video cross-modal recommendation, i.e. recommending a music track for a video or vice versa. We rely on a self-supervised learning paradigm to learn from a large amount of unlabelled data. We rely on a self-supervised learning paradigm to learn from a large amount of unlabelled data. More precisely, we jointly learn audio and video embeddings by using their co-occurrence in music-video clips. In this work, we build upon a recent video-music retrieval system (the VM-NET), which originally relies on an audio representation obtained by a set of statistics computed over handcrafted features. We demonstrate here that using audio representation learning such as the audio embeddings provided by the pre-trained MuSimNet, OpenL3, MusicCNN or by AudioSet, largely improves recommendations. We also validate the use of the cross-modal triplet loss originally proposed in the VM-NET compared to the binary cross-entropy loss commonly used in self-supervised learning. We perform all our experiments using the Music Video Dataset (MVD).

Since creating labelled data in the field of remote sensing requires time and manpower, it has become important to use unlabelled data. In this paper we study a semi supervised long short term memory autocoder approach for crop classification with multi-temporal remote sensing data. In this study, a long short-term memory autoencoder network was trained with unlabelled data and the learned weights were used for the initialization of a model trained for classification with labelled data. The challenging Breizhcrops time series dataset, and multitemporal images of the Sakarya region were used for validation. It has been observed that the network trained with unlabelled data outperformed the network trained with only labelled data. The results shed light on the potential for unlabelled data use in the crop classification field.

Subspace clustering is a popular method for clustering unlabelled data. However, the computational cost of the subspace clustering algorithm can be unaffordable when dealing with a large data set. Using a set of dimension sketched data instead of the original data set can be helpful for mitigating the computational burden. Thus, finding a way for dimension sketching becomes an important problem. In this paper, a new dimension sketching algorithm is proposed, which aims to select informative dimensions that have significant effects on the clustering results. Experimental results reveal that this method can significantly improve subspace clustering performance on both synthetic and real-world datasets, in comparison with two baseline methods.

Machine Learning (ML) and Deep Learning (DL) techniques have achieved remarkable results in many domains resulting in excellent performances in fields including Natural Language Processing (NLP), Computer Vision (CV) and Speech Recognition (SR). In light of these achievements, the application of ML algorithms to the industry world is appealing and very interesting results have already been produced by the scientific community. However, industrial data are often significantly different from those commonly employed in the disciplines where DL has had a larger impact. Industrial datasets are often limited in size and are constituted by large amounts of unlabelled data. In contrast, in CV and NLP, large deep neural networks can benefit from large labelled datasets, that can thus be used to train DL models in a fully-supervised fashion. In this work, we adopt a self-supervised approach to process heterogeneous dataset characterized by a relatively large number of unlabelled data and few labelled ones. We apply the proposed approach to two industrial time-series data and show that leveraging unlabelled data results in improved performance on classification downstream tasks.

2021 24th Euromicro Conference on Digital System Design (DSD) : [Proceedings] - IEEE, 2021. - ISBN 978-1-6654-2703-6 - doi:10.1109/DSD53832.2021.00028 24th Euromicro Conference on Digital System Design (DSD), Palermo, Italy, 1 Sep 2021 - 3 Sep 2021; 5 pp. (2021). doi:10.1109/DSD53832.2021.00028, Anomaly detection algorithms solve the problem of identifying unexpected values in data sets. Such algorithms have been classically used for cleaning unlabelled data sets from potentially unwanted values. However, the ability to detect outlying values in data sets can also be used to detect anomalies in systems. Semi-supervised anomaly detection algorithms learn from data for known correct behavior. Such algorithms have been used in various fields, e.g., system security, fault detection, medical applications.In this paper, we use the Area Under the Receiver Operating Characteristic (AUROC) score to evaluate algorithms for semi-supervised anomaly detection when applied to high-integrity distributed digital systems. We identify the relevant parameter for each algorithm and observe how the parameter influences the score and the runtime.

The Internet of Things (IoT) is a network of distributed devices or sensors connected through the internet to allow gathering and sharing of data. The data generated by these devices is affected by anomalies or abnormal behaviour due to attack issues, or breakdown in devices, as examples. However, most current anomaly detection systems rely on labelled data, while the class labels for IoT data are usually unavailable. Furthermore, the manual labelling task is expensive and time-consuming to perform due to the need for domain experts. More importantly, the volume of data in the IoT is growing rapidly, creating a need to predict the classification labels for future data. This study proposes a Hybrid Learning Model which uses both Clustering and Classification methods (HLMCC) to automate the labelling process and detect anomalies in IoT data. The model consists of two practical phases, automatic labelling and detecting anomalies. First, the HLMCC groups the data into normal and anomaly clusters by adopting Hierarchical Affinity Propagation (HAP) clustering. Second, the labelled data obtained from the clustering phase is used to train the Decision Trees (DTs) and to classify future unseen data. The results show that the HLMCC is able to automate the labelling of data, which is beneficial to minimize human involvement. Moreover, HLMCC outperforms the DTs on the originally labelled datasets and the state-of-the-art model over a wide range of evaluation metrics based on the average ranks. HLMCC produces the highest average ranks against other models in terms of False Positive Rate (FPR), recall, precision and the Area Under the Precision-Recall curve (AUCPR) with 1.8, 1.6, 1.8 and 1.8, respectively.

In the past decade, a plethora of efforts were given to the field of facilitating a better smart grid system by leveraging the power of artificial intelligence. Undoubtedly, machine learning is currently playing an increasingly important role in almost every aspect of power systems. However, in real practice, there is a much larger amount of unlabelled data than the one labelled by human experts. In this work, we make a perspective study on overcoming the data scarcity in smart grid. The active learning strategy will be proposed to provide a feasible solution for addressing the data scarcity challenge. In addition, we will give a discussion on current state-of-the-art and the limitations in previous work. We hope this work can be a good guide for researchers to further the relevant study in the near future.

In this paper, we present a novel image clustering method that converts a typical image clustering problem into a graph node classification task, given only unlabelled data samples. Graph convolutional network is utilized to perform graph representation encoding. Unlike traditional clustering approaches relying on hand-crafted criteria, our method learns to cluster graph-structured data by maximizing mutual information between the global-graph representation and local-graph representations. The learnt graph embeddings can preserve global information in all locations of local-graphs such as pseudo labels for image clustering tasks. To test the generalization and robustness of the proposed method, we conduct experiments on two different benchmarks, such as fashion classification and finegrained object classification. The preliminary experiment results show that the proposed method outperforms all the baselines.

It is not uncommon for data obtained from systems to change after the model is learned. These occurrences are named dataset shifts and to deal with them models with the ability to adapt to data changes must be used. A strategy that can be easily integrated to other classifiers is proposed. It creates a geometrical representation of data that extracts information from both labelled and unlabelled data. Then data entropy and Jensen-Shannon dissimilarity tests are used during the model selection to handle cases where data shift. Results have shown that the proposed method is promising because of its simple integration with state of art classifiers and its performance in enhancing said classifiers accuracy in the studied cases.

In supervised medical image learning, labels for disease images are indispensable. The correctness of labels has a significant influence on disease discrimination. However, the labelling of disease images by professional doctors is a time-consuming and labour-intensive project; therefore, obtaining correct and high-quality labels is a difficult task in medical image learning. This paper proposes a deep convolutional neural network-based label completion and correction strategy for supervised medical image learning. We take the diagnosis of Seborrheic Keratosis (SK) and Flat Wart (FW) as examples and perform the label completion and correction on a dataset containing confocal laser scanning microscope images. Experimental results show that this strategy can use limited labelled data to complete the labels for most of the unlabelled data and correct some noise labels in the dataset, which improves the accurate rate of the model to identify diseases.

Fault diagnosis plays a crucial role to maintain healthy conditions in rotating machinery. This paper proposes a framework to detect new patterns of abnormal conditions in gearboxes, that would be associated to new faults. This is achieved through a Hybrid Heuristic Algorithm for Evolving Models in scenarios of Classification and Clustering (HHA-EMCC), which is a machine learning algorithm that can be adapted to solve problems related to classification and clustering both combined. The design aims at creating clusters and classes inspired by the main principles of the nearest neighbour (1-NN) strategy and Kmeans. HHA-EMCC has the particularity of detecting new clusters after being trained, this characteristic defines some guidelines that determine whether a cluster represents new knowledge or not. The framework is able to discover abnormal conditions from unlabelled data through cluster constructions. This analysis can lead to labelling these clusters as new classes. Once a new pattern is identified, the associated data feeds the current classifier for a new training phase. The proposed framework is tested on a fault dataset for gearboxes and experimental results show that valuable new knowledge is obtained.

We propose a fast progressive transductive support vector machine (FPTSVM) method for detecting changes in bitemporal multispectral images. Our method differs from other transductive SVM methods in that much fewer unlabelled data vectors need to be considered in each iteration for the determination of transductive samples. As a result, the computational cost of the method is reduced. To improve the change detection results, our method incorporates the spectral angle (SA) as an extra component to each input data vector. Experimental results on Landsat dataset show that our method gives the smallest overall error and the fastest run time.

Anomaly detection comprises the identification of observations which do not follow the expected patterns of the assumed data set. We attempt to simplify the problem of textual anomaly detection by constructing a Multinomial Naive Bayes classifier and enhancing it with an augmented Expectation Maximization (EM) algorithm. By doing so, we utilize large amounts of unlabelled data and show how the EM algorithm could increase the accuracy of the Naive Bayes classifier. The process is applied to a binary classification environment in order to detect anomalies in text.

The problem of utilizing machine learning approach in Indonesian Named Entity Recognition (NER) system is the limited amount of labelled data for training process. However, unlike the limited availability of labelled data, unlabelled data is widely available from many sources. This enables a semi-supervised learning approach to solve this NER system problem. This research aims to design a semi-supervised learning model to solve NER system problem. A semi-supervised co-training learning is used to utilize unlabelled data in NER learning process to produce new labelled data that can be applied to enhance a new NER classi□cation system. This research uses two kinds of data, Indonesian DBPedia data as labelled data and news article text from Indonesian news sites (kompas.com, cnnindonesia.com, tempo.co, merdeka.com and viva.co.id) as unlabelled data. The pre-processing steps applied to analyze unstructured text are sentence segmentation, tokenization, stemming, and PoS Tagging. The results of this pre-process are the NER and its context used as unlabelled data for the semi-supervised co-training process. The SVM algorithm is used as a classi□cation algorithm in this process. 10 Cross Fold Validation is used as the system testing approach. Based on the result of the NER testing system, the precision is 73.6%, the recall is 80.1% and f1 mean is 76.5%.

In recent years, flow-based anomaly detection has been used as a scalable method for high-speed networks. The application of supervised learning in flow-based anomaly detection has been considered in a number of studies. However, supervised methods are not very useful as they are only trained with labelled data. Therefore, in this study, we use a semi-supervised method to address the problem of the limitation of labelled data. S4VM is a semi-supervised method which can work with both labelled and unlabelled data. This method is used in this study to detect anomalies in flow traffic. The results show that S4VM has high accuracy when a large proportion of data is unlabelled. Although the accuracy of S4VM is slightly less than supervised learning, this method reduces the cost of labeling data, as only a small number of labelled flows are required. To evaluate the proposed anomaly detection method, a number of flow-based datasets are generated.

Internet communities have become an important source of support for people with chronic illnesses such as diabetes and obesity, both of which have been associated with depression. In this paper, we argue text classification as promising tool for mining mood disorder cues from Internet chat messages. We created a minimal corpus of 200 chat profiles, based on a disease classification system, ICD-10 diagnostic criteria, and DSM-IV depression definitions. Using significant grams, we trained and tested multiple classifiers on the corpus, with additional evaluation on unlabelled data. Current findings demonstrate the feasibility of scalable flagging of patients who areat risk of developing severe depression in large Internet health communities.

Active Learning is a technique to automatically select the useful instances from the unlabelled data in such a way that, when these are augmented to the training data, overall classification performance improves. The creation of training examples otherwise involves significant amount of costs and efforts and hence, is a major constraint in the supervised algorithms. In this paper, we investigate the effectiveness of active learning for sentiment classification of Tweets. The algorithm selects the informative unlabelled data based on the concept of uncertainty sampling which dictates that only those Tweets be added to the training set for which the classifier can quickly refine its decision boundary. Our experiments on a benchmark dataset of Tweets show an overall accuracy of 83.95%, which is an increment of 6.75% over the baseline model, constructed by training a Support Vector Machine (SVM) with all the available set of features. The approach, being very general, is scalable, domain-adaptable and easy to implement for a wide variety of problems.

Training a statistical surface realiser typically relies on labelled training data or parallel data sets, such as corpora of paraphrases. The procedure for obtaining such data for new domains is not only time-consuming, but it also restricts the incorporation of new semantic slots during an interaction, i.e. using an online learning scenario for automatically extended domains. Here, we present an alternative approach to statistical surface realisation from unlabelled data through automatic semantic slot labelling. The essence of our algorithm is to cluster clauses based on a similarity function that combines lexical and semantic information. Annotations need to be reliable enough to be utilised within a spoken dialogue system. We compare different similarity functions and evaluate our surface realiser—trained from unlabelled data—in a human rating study. Results confirm that a surface realiser trained from automatic slot labels can lead to outputs of comparable quality to outputs trained from human-labelled inputs.

In many data mining and machine learning applications, data may be easy to collect. However, labeling the data is often expensive, time consuming or difficult. Such applications give rise to semi-supervised learning techniques that combine the use of labelled and unlabelled data. Co-training is a popular semi-supervised learning algorithm that depends on splitting the features of a data set into two redundant and independent views. In many cases however such sets of features are not naturally present in the data or are unknown. In this paper we test feature splitting methods based on maximizing the confidence and the diversity of the views using genetic algorithms, and compare their performance against random splits. We also propose a new criterion that maximizes the complementary nature of the views. Experimental results on six different data sets show that our optimized splits enhance the performance of co-training over random splits and that the complementary split outperforms the confidence, diversity and random splits.

Question classification is crucial for the automatically question answering. And Random Walk is a promising approach for semi-supervised learning problems of learning from labeled and unlabeled data. Given a set of points, some of them are labeled, and the remaining points are unlabeled, the goal is to predict the labels of the unlabeled points. Since labeling often requires expensive human labor, whereas unlabelled data is easier to obtain, semi-supervised learning is very useful in many real-world problems, such as text classification. Here we proposed an approach for Chinese question Classification using Multilevel Random Walk (MRK), which is an improvement of random walk. In this paper, we selected four kinds of features (words, pos, named entity, semantic) to present Chinese questions, and carried out experiments to validate the method on a large-scale real-world dataset.

In many important image classification problems, acquiring class labels for training instances is costly, while gathering large quantities of unlabelled data is cheap. A semiautomated system for the classification of bronchovascular pairs based on co-training in high resolution computed tomography (HRCT) images is presented. A bronchovascular pair is formed between a bronchus and a vessel. The identification of such structures provides valuable diagnostic information in patients with suspected airway diseases. Co-training is a semi-supervised multi-view learning algorithm where classifiers trained with a small number of labelled examples are improved by augmenting the small training set with a large pool of unseen examples. We incorporate active learning where the user labels examples on which the two views disagree. The two views in our system are based on spatial relations and ERS, a gradient based feature set. In addition, the optimal parameters required in the pre-processing step before feature extraction and recognition was automatically chosen. The system was co-trained on 41 unlabelled HRCT scans selected from 26 patient studies. It was successfully evaluated on 26 other HRCT scans manually labelled in consultation with radiologists.