Computer vision is a major branch of artificial intelligence algorithm. The algorithm of computer vision mainly consists of the processing of image and video, including image recognition and image detection etc. Pract...
详细信息
ISBN:
(数字)9781728147437
ISBN:
(纸本)9781728147444
Computer vision is a major branch of artificial intelligence algorithm. The algorithm of computer vision mainly consists of the processing of image and video, including image recognition and image detection etc. Practice has proved that computer vision is scientific and practical to a certain extent. In pace with the development of in-depth learning, computer vision has already been put to use well in all walks of life. However, it is still in exploring stage in the medical field, because the medical data is sensitive, which requires high accuracy of the algorithm. In this paper, images of PCam medical electron microscope are put to use for tumor detection, which is an task of image recognition and an automatic encoder is used to lower the dimensions of the data into low-dimensional vectors which are used as features in training. Then the vectors are added as features to the training, and the model is trained together with the original data set as the training features of NASnet. Because detection algorithms in the medical field pay more importance to the true positive rate and false positive rate. When the output is positive, it is necessary to be revalidated by SVM model trained by encoder. As a result, ROC curve is 0.98, which is 0.03 higher than Baseline.
Crop classification maps based on high resolution remote sensing data are essential for supporting sustainable land management. The most challenging problems for their producing are collecting of ground based training...
详细信息
ISBN:
(纸本)9781538671504
Crop classification maps based on high resolution remote sensing data are essential for supporting sustainable land management. The most challenging problems for their producing are collecting of ground based training and validation datasets, non-regular satellite data acquisition and cloudiness. To increase the efficiency of ground data utilization it is important to develop classifiers able to be trained on the data collected in the previous year. In this study, we propose new deep learning method for providing crop classification maps using in-situ data that has been collected in the previous year. Main idea of the study is to utilize deep learning approach based on sparse autoencoder. At the first stage it is trained on satellite data only and then neural network fine-tuning is conducted based on in-situ data form the previous year. Taking into account that collecting ground truth data is very time consuming and challenging task, the proposed approach allows us to avoid necessity for annual collecting in-situ data for the same territory. Experimental results for the territory of Ukraine show that this technique is rather efficient and provides reliable crop classification maps with overall accuracy higher than 85.9%.
In this paper, we present a deep learning based modulation scheme for chaotic orthogonal frequency division multiplex (OFDM) transmissions over non-contiguous frequency bands of cognitive radio systems. In cognitive r...
详细信息
ISBN:
(纸本)9781538661192
In this paper, we present a deep learning based modulation scheme for chaotic orthogonal frequency division multiplex (OFDM) transmissions over non-contiguous frequency bands of cognitive radio systems. In cognitive radio systems, the users access the spectrum bands dynamically and the corresponding channel characteristics also changes. Different from the traditional modulation scheme that uses the fixed mapping pattern to modulate the signals, we propose to apply the deep learning method to build up the constellations intelligently. Based on the autoencoder architecture of deep learning, we construct the constellation mapping and demapping patterns adaptively with the aim to minimize the bit error rate (BER) over the dynamically changing non-contiguous channels. Simulation results over additive white Gaussian noise (AWGN) channel and Rayleigh fading channel show that our proposed system achieves better BER performances for legitimate receivers when compared with the conventional modulation schemes. In addition, the presented scheme remains the high security performance thanks to the usage of the chaotic sequences.
To provide reliable crop maps for the same territory each year, it is necessary to collect in-situ data for each year independently. Collecting ground truth data is a very time consuming and challenging task. At prese...
详细信息
ISBN:
(纸本)9781538663837
To provide reliable crop maps for the same territory each year, it is necessary to collect in-situ data for each year independently. Collecting ground truth data is a very time consuming and challenging task. At present, unfortunately, there is no an adopted approach, how to utilize in-situ and satellite data from previous years for crop mapping in the subsequent years. In this paper, we propose a new deep learning approach using sparse autoencoder based on only satellite data, and a further procedure of neural network fine-tuning based on in-situ data. The possibility of utilizing this deep learning architecture based on translating all available satellite data into the unified hyperspace. The study is carried out for the central part of Ukraine. Obtained results show that this technique is feasible and provides reliable crop classification maps with overall accuracy (OA) of 91.0% and 85.9% for two different experiments. The use of the proposed approach makes it possible to avoid, or decrease, the necessity for collecting in-situ data for each year and for each part of large territory.
This paper proposes innovative anomaly detection technologies for manufacturing systems. We combine the event ordering relationship based structuring technique and the deep neural networks to develop the structured ne...
详细信息
ISBN:
(纸本)9781538673454
This paper proposes innovative anomaly detection technologies for manufacturing systems. We combine the event ordering relationship based structuring technique and the deep neural networks to develop the structured neural networks for anomaly detection. The event ordering relationship based neural network structuring process is performed before neural network training process and determines important neuron connections and weight initialization. It reduces the complexity of the neural networks and can improve anomaly detection accuracy. The structured time delay neural network (TDNN) is introduced for anomaly detection via supervised learning. To detect anomaly through unsupervised learning, we propose the structured autoencoder. The proposed structured neural networks outperform the unstructured neural networks in terms of anomaly detection accuracy and can reduce test error by 20%. Compared with popular methods such as one-class SVM, decision trees, and distance-based algorithms, our structured neural networks can reduce anomaly detection misclassification error by as much as 64%.
The exponential growth of the data collected by telescopes have turned astronomy into a data-drive science. The detection of astronomical transient events, short-lived and bright phenomena such as the Supernovae, is c...
详细信息
ISBN:
(纸本)9781509060146
The exponential growth of the data collected by telescopes have turned astronomy into a data-drive science. The detection of astronomical transient events, short-lived and bright phenomena such as the Supernovae, is currently a main science driver of many astronomical surveys. There is an opportunity for the application of machine learning methods for the automatic detection of astronomical transients. In this paper we focus on the unsupervised learning case to perform an exploratory analysis on a dataset of 1,250,000 astronomical transient candidates from the High Cadence Transient Survey. Our contributions can be summarized in 1) The application of Deep Variational Embedding for latent space clustering of a large database of transient candidates obtaining a clustering accuracy of 95:33% and 2) The proposal of an auto-regularization term as a novel approach to solve the common problem of over-regularization in variational autoencoders, we show that using this term not only improves the convergence of the algorithm but also increases the clustering accuracy and reconstruction quality.
Deep learning has the potential to drastically increase the accuracy and efficiency of prostate cancer diagnosis, which would be of uttermost use. Today the diagnosis is determined manually from H&E stained specim...
详细信息
ISBN:
(纸本)9781538636367
Deep learning has the potential to drastically increase the accuracy and efficiency of prostate cancer diagnosis, which would be of uttermost use. Today the diagnosis is determined manually from H&E stained specimens using a light microscope. In this paper several different approaches based on convolutional neural networks for prostate cancer classification are presented and compared, using three different datasets with different origins. The issue that algorithms trained on a certain site might not generalize to other sites, due to for example inevitable stain variations, is highlighted. Two different techniques to overcome this complication are compared;by training the networks using color augmentation and by using digital stain separation. Furthermore, the potential of using an autoencoder to get a more efficient downsampling is investigated, which turned out to be the method giving the best generalization. We achieve accuracies of 95% for classification of benign versus malignant tissue and 81% for Gleason grading for data from the same site as the training data. The corresponding accuracies for images from other sites are in average 88% and 52% respectively.
In this paper, we present a deep learning based approach to assign POS tags to words in a piece of text given to it as input. We propose an unsupervised approach owing to the lack of a large Sanskrit annotated corpora...
详细信息
ISBN:
(纸本)9781450366250
In this paper, we present a deep learning based approach to assign POS tags to words in a piece of text given to it as input. We propose an unsupervised approach owing to the lack of a large Sanskrit annotated corpora and use the untagged Sanskrit Corpus prepared by JNU for our purpose. The only tagged corpora for Sanskrit is created by JNU which has 115,000 words which are not sufficient to apply supervised deep learning approaches. For the tag assignment purpose and determining model accuracy, we utilize this tagged corpus. We explore various methods through which each Sanskrit word can be represented as a point multi-dimensional vector space whose position accurately captures its meaning and semantic information associated with it. We also explore other data sources to improve performance and robustness of the vector representations. We use these rich vector representations and explore autoencoder based approaches for dimensionality reduction to compress these into encodings which are suitable for clustering in the vector space. We experiment with different dimensions of these compressed representations and present one which was found to offer the best clustering performance. For modelling the sequence in order to preserve the semantic information we feed these embeddings to a bidirectional LSTM autoencoder. We assign a POS tag to each of the obtained clusters and produce our result by testing the model on the tagged corpus.
Radar Cross Section (RCS) is a measurement of scattering performance of an object. RCS plays an important role in the design of stealth weapon system. However, the method of calculating RCS of every angle of azimuth a...
详细信息
ISBN:
(纸本)9781538663967
Radar Cross Section (RCS) is a measurement of scattering performance of an object. RCS plays an important role in the design of stealth weapon system. However, the method of calculating RCS of every angle of azimuth and elevation is complex, and the efficiency is low. Evidently, it is of great significance for predicting the unknown RCS given a set of RCS data of an object. Towards this aim, we propose a novel approach using multilayers Long Short Term Memory (LSTM) networks based on unsupervised learning - the mechanism is similar to that of autoencoder. The encoder LSTM maps the input RCS into a fixed length representation. The decoder LSTM decodes the representation to predict RCS. For the purpose of this work, we create an 3D object and collect its RCS for a dataset with the help of electromagnetic simulation software, FEKO. The proposed networks have been applied on the test dataset and yield satisfactory results.
The present paper aims to propose a new type of learning method to increase information content in input patterns with multiple steps to be used in supervised learning. Unsupervised pre-training to train multi-layered...
详细信息
ISBN:
(纸本)9781538674475
The present paper aims to propose a new type of learning method to increase information content in input patterns with multiple steps to be used in supervised learning. Unsupervised pre-training to train multi-layered neural networks turned out to be not so effective as has been expected, because connection weights obtained by the unsupervised learning tend to lose their original characteristics immediately in supervised training. To keep original information by unsupervised learning, we here try to increase information in input patterns as much as possible to overcome the vanishing information problem. In particular, for acquiring detailed information more appropriately, we gradually increases detailed information through multiple steps. We applied the method to the actual real data set of the eye-tracking, and two step information augmentation approach was taken. The results confirmed that generalization performance could be improved. In addition, we could interpret the importance of input variables more easily by treating all connection weights collectively.
暂无评论