Considering that Internet-of-Things (IoT) devices are often deployed in highly dynamic environments, mainly due to their continuous exposure to end-users' living environments, it is imperative that the devices can...
详细信息
ISBN:
(数字)9783030342555
ISBN:
(纸本)9783030342555;9783030342548
Considering that Internet-of-Things (IoT) devices are often deployed in highly dynamic environments, mainly due to their continuous exposure to end-users' living environments, it is imperative that the devices can continually learn new concepts from data stream without catastrophic forgetting. Although simply replaying all the previous training samples can alleviate this catastrophic forgetting problem, it not only may pose privacy risks, but also may require huge computing and memory resources, which makes this solution infeasible for resource-constrained IoT devices. In this paper, we propose IL4IoT, a lightweight framework for incremental learning for IoT devices. The framework consists of two cooperative parts: a continually updated knowledge-base and a task-solving model. Through this framework, we can achieve incremental learning while alleviating the catastrophic forgetting issue, without sacrificing privacy-protection and computing-resource efficiency. Our experiments on MNIST dataset and SDA dataset demonstrate the effectiveness and efficiency of our approach.
Collaborative Filtering to Supervised Learning (COFILS) transforms a Collaborative Filtering (CF) problem into classical Supervised Learning (SL) problem. Applying COFILS reduces data sparsity and makes it possible to...
详细信息
Collaborative Filtering to Supervised Learning (COFILS) transforms a Collaborative Filtering (CF) problem into classical Supervised Learning (SL) problem. Applying COFILS reduces data sparsity and makes it possible to test a variety of SL algorithms rather than matrix decomposition methods. Its main steps are: extraction, mapping and prediction. Firstly, a Singular Value Decomposition (SVD) generates a set of latent variables from a ratings matrix. Next, on the mapping phase, a new data set is generated where each sample contains a set of latent variables from a user and each rated item;and a target that corresponds the user rating for that item. Finally, on the last phase, a SL algorithm is applied. One problem of COFILS is its dependency on SVD, that is not able to extract non-linear features from data and it is not robust to noisy data. To address this problem, we propose switching SVD to a Stacked Denoising Auto-encoder (SDA) on the first phase of COFILS. With SDA, more useful and complex representations can be learned in a neural network with a local denoising criterion. We test our novel technique, namely Auto-encoder COFILS (A-COFILS), on MovieLens, R3 Yahoo! Music and Movie Tweetings data sets and compare to COFILS, as a baseline, and state of the art CF techniques. Our results indicate that A- COFILS outperforms COFILS for all the data sets and with an improvement up to 5.9%. Also, A-COFILS achieves the best result for the MovieLens 100k data set and ranks on the top three algorithms for these data sets. Thus, we show that our technique represents an advance on COFILS methodology, improving its results and making it a suitable method for CF problem. (C) 2017 Elsevier Ltd. All rights reserved.
Purpose of ReviewMachine learning methods are increasingly used in health data mining. We describe current unsupervised learning methods for phenotyping and discovery and illustrate their application for detecting fea...
详细信息
Purpose of ReviewMachine learning methods are increasingly used in health data mining. We describe current unsupervised learning methods for phenotyping and discovery and illustrate their application for detecting features and sub-groups related to drug use within a *** FindingsPatient representation or phenotyping and discovery is one of the main branches of health data analysis. Phenotyping concerns identifying features that are representative of the population from raw patient data. Discovery involves analysing these features, for example, to identify patterns in the population such as sub-groups and to predict outcomes. Most studies use unsupervised learning methods for phenotyping as they are suited for data-driven feature extraction. We describe some of the commonly used methods and demonstrate their use in feature selection followed by cluster *** learning methods can be used to extract the features of and identify sub-groups within specific populations. We demonstrate the potential of these methods and highlight the associated challenges, which researchers may find useful in understanding the suitability of these methods for analysing health data.
Conventional models for emotion recognition from speech signal are trained in supervised fashion using speech utterances with emotion labels. In this study we hypothesize that speech signal depends on multiple latent ...
详细信息
ISBN:
(纸本)9781479981311
Conventional models for emotion recognition from speech signal are trained in supervised fashion using speech utterances with emotion labels. In this study we hypothesize that speech signal depends on multiple latent variables including the emotional state, age, gender, and speech content. We propose an Adversarial autoencoder (AAE) to perform variational inference over the latent variables and reconstruct the input feature representations. Reconstruction of feature representations is used as an auxiliary task to aid the primary emotion recognition task. Experiments on the IEMOCAP dataset demonstrate that the auxiliary learning tasks improve emotion classification accuracy compared to a baseline supervised classifier. Further, we demonstrate that the proposed learning approach can be used for the end-to-end speech emotion recognition, as its applicable for models that operate on frame-level inputs.
Zero-shot learning (ZSL) aims at recognizing unseen classes with knowledge transferred from seen classes. This is typically achieved by exploiting a semantic feature space (FS) shared by both seen and unseen classes, ...
详细信息
ISBN:
(纸本)9781538695524
Zero-shot learning (ZSL) aims at recognizing unseen classes with knowledge transferred from seen classes. This is typically achieved by exploiting a semantic feature space (FS) shared by both seen and unseen classes, i.e., attributes or word vectors, as the bridge. However, due to the mutually disjoint of training (seen) and testing (unseen) data, existing ZSL methods easily and commonly suffer from the domain shift problem. To address this issue, we propose a novel model called AMS-SFE. It considers the Alignment of Manifold Structures by Semantic Feature Expansion. Specifically, we build up an autoencoder based model to expand the semantic features and joint with an alignment to an embedded manifold extracted from the visual FS of data. It is the first attempt to align these two FSs by way of expanding semantic features. Extensive experiments show the remarkable performance improvement of our model compared with other existing methods.
We consider a smart phone scenario with a number of apps used by a user. The app usage data provides information about the user behavior, which can be used to identify the user demographics and interest and in turn is...
详细信息
ISBN:
(纸本)9783030327859;9783030327842
We consider a smart phone scenario with a number of apps used by a user. The app usage data provides information about the user behavior, which can be used to identify the user demographics and interest and in turn is used to find similar users. In this paper, we propose a method to generate a latent space user embedding using the user app usage data, which is a dense low-dimensional representation of the user. This representation is used for low latency user similarity computation and acts as the user feature representation in user demographics prediction models.
Gas metal arc welding is widely used in industrial series production for joining aluminum. A lot of factors, such as instabilities and complex dependencies, influence the quality of the resulting welding seams. It is ...
详细信息
Gas metal arc welding is widely used in industrial series production for joining aluminum. A lot of factors, such as instabilities and complex dependencies, influence the quality of the resulting welding seams. It is challenging to identify the causes of welding defects, and the real reason is not always well understood. Ensuring the process stability helps production workers to increase the overall production efficiency. The process stability increases the process repeatability, so the welding performance is optimized and rejects are avoided. This paper presents a technique to detect process instabilities within the multivariate process variables automatically. An autoencoder architecture is implemented. The latent space of the autoencoder and reconstruction of the time series are used to detect process instabilities. Detected issues are visualized in a heatmap, including supportive metrics to describe deviations from the expected behavior. As a result, the proposed architecture supports process optimization and leads to an increase in production transparency.
A smile is a specific movement of face muscles to relay an optimistic feeling. A smile represents satisfaction and happiness. Many application created using smile detection technology, for example product rating, pati...
详细信息
A smile is a specific movement of face muscles to relay an optimistic feeling. A smile represents satisfaction and happiness. Many application created using smile detection technology, for example product rating, patient monitoring, image capturing, video conferencing and interactive systems. There are many smile detections techniques have been proposed for smile detection in the unconstrained scenarios. However, the dimensions of most notable feature descriptors are humongous, which is challenging in real-time applications. Besides, feature should be more powerful to identify between smiling and non-smiling face. The proposed method has two consecutive actions: 1) amalgamation of geometric feature extraction (GFE) and regional local binary pattern (LBP) features extraction using autoencoders; 2) Kohonen selforganizing map (KSOM) is adopted to classify smile based on these features. The proposed method is mathematics more dynamic and performance wise more precise. The performance of the propounded approach is proved on GENKI-4K database.
A Network Intrusion Detection System is a critical component of every internet-connected system due to likely attacks from both external and internal sources. Such Security systems are used to detect network born atta...
详细信息
ISBN:
(纸本)9781728111513
A Network Intrusion Detection System is a critical component of every internet-connected system due to likely attacks from both external and internal sources. Such Security systems are used to detect network born attacks such as flooding, denial of service attacks, malware, and twin-evil intruders that are operating within the system. Neural networks have become an increasingly popular solution for network intrusion detection. Their capability of learning complex patterns and behaviors make them a suitable solution for differentiating between normal traffic and network attacks. In this paper, we have applied a deep autoencoded dense neural network algorithm for detecting intrusion or attacks in 5G and IoT network. We evaluated the algorithm with the benchmark Aegean Wi-Fi Intrusion dataset. Our results showed an excellent performance with an overall detection accuracy of 99.9% for Flooding, Impersonation and Injection type of attacks. We also presented a comparison with recent approaches used in literature which showed a substantial improvement in terms of accuracy and speed of detection with the proposed algorithm.
General named entity recognition systems exclusively focus on higher accuracy regardless of dirty data. However, raw source data face serious challenges specially that are originated from automated speech recognition ...
详细信息
ISBN:
(纸本)9783030299088;9783030299071
General named entity recognition systems exclusively focus on higher accuracy regardless of dirty data. However, raw source data face serious challenges specially that are originated from automated speech recognition systems' results. In this paper, we propose Pinyin (Pinyin is the official romanization system for Standard Chinese, each Chinese character has its own pinyin sequence which is composed of Latin alphabet) Hierarchical Attention Encoder-Decoder network and Character Alternate Network to overcome Chinese homophones' problems which frequently frustrate researchers in consecutive Natural Language Understanding (NLU). Our models present a none word segmentation structure to effectively avoid secondary data corruption and adequately extract words' internal features. Besides, corrupted sequences can be revised by character-level network. Evaluation demonstrates that our proposed method achieves 93.73% F1 scores which are higher than 90.97% F1 scores using baseline models in homophone-noisy dataset. Additional experiments are conducted to show equivalent results in the universal dataset.
暂无评论