As the streaming data generated by Internet of Things (IoT) ubiquitous sensors grow in massive scale, extracting interesting information (anomalies) in real-time becomes more challenging. Traditional systems which ret...
详细信息
ISBN:
(纸本)9781728145693
As the streaming data generated by Internet of Things (IoT) ubiquitous sensors grow in massive scale, extracting interesting information (anomalies) in real-time becomes more challenging. Traditional systems which retrospectively perform all the processing in the cloud do not capture real-time changes in the data. Similarly, real-time solutions which rely on human monitors have the tendency to miss the anomalies due to their rare nature. In recent times, several machine learning techniques have been proposed for stream processing. Approaches based on supervised or semi-supervised learning fail to adapt to changing patterns of the streaming data and the data labelling costs are huge. To address these limitations, we propose a cloud-assisted framework where an intermediary node (edge) is introduced between the end devices and the cloud to assist in stream processing. A model deployed on the edge is designed to learn in an iterative manner to discriminate between similar and dissimilar data representations, making it easier to distinguish the anomalies. In this work, we have proposed an iterative method that combines the capabilities of deep clustering and l(2)-normalisation to achieve better discriminative representations. Experimental results demonstrate the proposed method achieves robust performance over state-of-the-art discriminative representation algorithms and sets new benchmark accuracy on transformation invariant image dataset.
Many fields of science and engineering require the use of complex and computationally expensive models to understand the involved processes in the system of interest. Nevertheless, due to the high cost involved, the r...
ISBN:
(数字)9781509066315
ISBN:
(纸本)9781509066322
Many fields of science and engineering require the use of complex and computationally expensive models to understand the involved processes in the system of interest. Nevertheless, due to the high cost involved, the required study becomes a cumbersome process. This paper introduces an interpolation procedure which belongs to the family of active learning algorithms, in order to construct cheap surrogate models of such costly complex systems. The proposed technique is sequential and adaptive, and is based on the optimization of a suitable acquisition function. We illustrate its efficiency in a toy example and for the construction of an emulator of an atmosphere modeling system.
The application of effective techniques like facial recognition and object detection in the domain of IoT has revolutionized the levels of control and accuracy sensors have had for a variety of purposes. Face recognit...
详细信息
ISBN:
(数字)9781728123394
ISBN:
(纸本)9781728123400
The application of effective techniques like facial recognition and object detection in the domain of IoT has revolutionized the levels of control and accuracy sensors have had for a variety of purposes. Face recognition modules have established their immense usefulness, and their impact has overweighed most other traditional forms of access control. It remains one of the most secure methods to authenticate human beings to grant them access to restricted areas or systems in an organization. However, this incomparable rise in accuracy comes with a relatively huge price. The algorithms used in facial recognition are complex mathematical functions that are iteratively performed on vast volumes of data continuously. In IoT applications, the embedded systems that run these algorithms are at all times clocked to their maximum processing power, they need not run in cases when there are no humans are present in the vicinity. This is inefficient, especially in systems that need to perform face recognition round-the-clock for authentication. This paper proposes a simple, intuitive and efficient solution to conserve the processor from being clocked to its maximum throughput without compromising on the high level of security that the face recognition algorithm offers.
Neural networks are used in many tasks today. One of them is the images processing. Autoencoder is very popular neural networks for such problems. Denoising autoencoder is an important autoencoder because some tasks w...
ISBN:
(数字)9781728199573
ISBN:
(纸本)9781728199580
Neural networks are used in many tasks today. One of them is the images processing. Autoencoder is very popular neural networks for such problems. Denoising autoencoder is an important autoencoder because some tasks we need a preprocessed image to get less noisy result. This research describes ways to analyze noisy images produced by a physically-based render engine and how to reduce that noise. The results showed that the algorithms are logarithmic.
Poor image quality in low light images may result in a reduced number of feature matching between images. In this paper, we investigate the performance of feature extraction algorithms in low light environments. To fi...
详细信息
ISBN:
(数字)9788993215205
ISBN:
(纸本)9781728185620
Poor image quality in low light images may result in a reduced number of feature matching between images. In this paper, we investigate the performance of feature extraction algorithms in low light environments. To find an optimal setting to retain feature matching performance in low light images, we look into the effect of changing feature acceptance threshold for feature detector and adding pre-processing in the form of Low Light image Enhancement (LLIE) prior to feature detection. We observe that even in low light images, feature matching using traditional hand-crafted feature detectors still performs reasonably well by lowering the threshold parameter. We also show that applying LLIE algorithms can improve feature matching even further when paired with the right feature extraction algorithm.
In this paper, the traffic sign recognition module of a small-scale autonomous car prototype will be presented. The process undergoing the choice of an appropriate algorithm, as well as the factors taken into consider...
详细信息
ISBN:
(纸本)9781728108780
In this paper, the traffic sign recognition module of a small-scale autonomous car prototype will be presented. The process undergoing the choice of an appropriate algorithm, as well as the factors taken into consideration will be presented in the form of a case study. The current literature presents various ways of achieving the recognition of traffic signs, but most of them are computational expensive, or have difficulty in offering consistent results in conditions that are different from the prerecorded ones. Since the processing on the car is carried on an embedded platform from Nvidia (Jetson TX2), this study is based on the same board, the stream being captured with a low-cost webcam. Classical algorithms like SURF (Speeded Up Robust Features), SIFT (Scale Invariant Feature Transform) or ORB (Oriented fast and Rotated Brief) offer reliable results when the lighting condition between the reference image and the image obtained from the camera are similar. In our setup, the algorithms mentioned above start to behave badly in low light conditions. Therefore, this paper discusses the possibility of using Haar like features alongside a classifier for detecting traffic signs.
image complexity has been estimated by computational algorithms that try to simulate the human criterion to determine complexity. image complexity has been used to study the behavior of the human brain, as well as in ...
详细信息
ISBN:
(纸本)9781538617281
image complexity has been estimated by computational algorithms that try to simulate the human criterion to determine complexity. image complexity has been used to study the behavior of the human brain, as well as in the area of imageprocessing. This paper presents a work to analyze image complexity. A fuzzy inference system using principal component analysis, PCA, is designed to model the human criterion. The PCA characteristics are obtained from the most used features reported in the literature;contrast, correlation, energy, homogeneity, frequency factor, edge density, compression ratio, number of regions, colorfulness, number of colors, and color harmony. The work was achieved with the data based RS1 and RS2. The results obtained by cross validation, demonstrate a correlation level of 0.8566 between the proposed method and the human criterion.
In recent years, with the development of machine learning technology, neural networks have gradually become a convenient method for classification of remote sensing image features. This article briefly describes the s...
详细信息
ISBN:
(数字)9781728183046
ISBN:
(纸本)9781728194608
In recent years, with the development of machine learning technology, neural networks have gradually become a convenient method for classification of remote sensing image features. This article briefly describes the structure and principle of the process of remote sensing image feature recognition, using three remote sensing image data sets AID, NWPU-RESISC45, UC Merced Land Use dataset for algorithm testing. First, the AlexNet neural network is used to extract the remote sensing image features, and the KNN is used to achieve image classification. The effects of extracting different alexnet feature layers on the average classification accuracy on the three data sets are compared. This paper compares the advantages of KNN in terms of time through PCA dimensionality reduction and k-means clustering optimization before classification, at the end of the article, it summarizes and briefly describes the development trend of neural network in the application of remote sensing image features classification technology.
Focusing on the Wavelet Transform, the paper explores four parallel Wavelet Transform algorithms and techniques from the perspectives of data parallel and algorithm parallel for remote sensing images. Among them, the ...
详细信息
ISBN:
(数字)9781728199283
ISBN:
(纸本)9781728199290
Focusing on the Wavelet Transform, the paper explores four parallel Wavelet Transform algorithms and techniques from the perspectives of data parallel and algorithm parallel for remote sensing images. Among them, the algorithm based on "Working Pool Parallel" achieves dynamic load balance without any limits to the scale of the data and the number of the Slaves. Therefore, this algorithm is easier to achieve the goal of processing the vast data of remote sensing images rapidly in the distributed network systems.
暂无评论