Loading...
Search for:
deep-learning
0.013 seconds
Total 304 records
Extraction and Processing Urban Data for Modeling Particulate Matter Concentrations in Tehran Using Probabilistic Neural Network
, M.Sc. Thesis Sharif University of Technology ; Arhami, Mohammad (Supervisor) ; Amini, Zahra (Co-Supervisor)
Abstract
The hourly concentrations of particulate matter in Tehran are modelled in this study. High levels of particles are one of the main air pollution challenges in this metropolis, especially in the colder seasons. A probabilistic neural network is used for modelling. The model uses Bayes' theorem which has a very high ability to tackle the complexities and uncertainties. Traffic, meteorology, land use, baseline concentration (at 5 am), vegetation, along with other data including the location of each station, time of recording each concentration data, area and population of the municipal district of each station are considered. This research introduced a cheap and accurate method for collecting...
Deep Learning for Speech Recognition
, M.Sc. Thesis Sharif University of Technology ; Sameti, Hossein (Supervisor)
Abstract
Speech recognition is one of the first goals of speech processing. Our goal in this thesis is to use deep learning for speech recognition. In recent years little improvement of speech recognition accuracies are reported. Deep learning is a new learning algorithm that results in improvement in many machine learning tasks. Following improvements reported in speech recognition in English language by deep learning, in this thesis we tried to improve accuracy over common and new recognition methods for Persian language.
First the overall structure of a typical speech recognition system is introduced. For this purpose, the modules of a speech recognition system are introduced. Deep multilayer...
First the overall structure of a typical speech recognition system is introduced. For this purpose, the modules of a speech recognition system are introduced. Deep multilayer...
Fine-grained Image Classification
, M.Sc. Thesis Sharif University of Technology ; Kasaei, Shohreh (Supervisor)
Abstract
Fine-grained image classification is image classification where the considered classes are all sub-classes of a certain, more general class. In this setting of the problem, the classes are visually very similar to each other, such that an unskilled human cannot discriminate between them. In this case, proposed methods for the ordinary image classification problem do not obtain good classification accuracy. So proposing new methods for solving this problem is necessary. In this thesis two new methods, based on recent advances in deep learning are proposed for solving the fine-grained image classification problem. First by improving several parts of one of the recent proposed methods for this...
Deep Learning Approach for Domain Adaptation
, M.Sc. Thesis Sharif University of Technology ; Soleymani Baghshah, Mahdieh (Supervisor)
Abstract
A predefined assumption in many learning algorithms is that the training and test data must be in thesame feature space and have the same distribution.However, this assumption may not hold in all of these algorithms and in the real world there might be difference between the source and the targer domian, whether in the feature space or the distribution. Moreover, there might be a few number of labled data of the target domain which causes difficulty in learning an accurate classifier. In such cases, transferring knowledge can be useful if can be done successfully and transfer learning was introduced for this purpose. Domain Adaptation is one of the transfer leaning problems that assume some...
Automatic Image Annotation Using Deep Learning
, M.Sc. Thesis Sharif University of Technology ; Jamzad, Mansour (Supervisor)
Abstract
With the advances in technology, Nowadays digital cameras are everywhere. As a result very large amount of images are on the web. Searching through these images intelligently and purposively is an essential need. Recently the possibility of retrieving images with some conceptual words along side of content based image retrieval has been studied in computer vision. For this purpose it’s required that for each image several words that describe its content be assigned automatically. One of the main problems for this task is semantic gap, meaning that the low level features such as color, texture,… don’t have the ability to describe the high level concepts in images which are comprehensible by...
Human Action Recognition in Smart Houses
, M.Sc. Thesis Sharif University of Technology ; Fatemizadeh, Emad (Supervisor)
Abstract
The definition of human action recognition is classification of input visual elements based on the action which is done by a person in the scene. One of the most important topics in the filed which has lots of applications is human action recognition in videos. Some of these applications are surveillance, video retrieval, human computer interaction and smart houses. Due to increments in number of alone elderly people, surveillance of them is one of the important applications of human action recognition. The challenges of the task are, camera movement, differences of environment and differences in acting by different actors.The goal of the project is proposing a deep convolutional neural...
Deep Zero-shot Learning
, M.Sc. Thesis Sharif University of Technology ; Soleymani, Mahdieh (Supervisor)
Abstract
In some of object recognition problems, labeled data may not be available for all categories. Zero-shot learning utilizes auxiliary information (also called signatures) describing each category in order to find a classifier that can recognize samples from categories with no labeled instance. On the other hand, with recent advances made by deep neural networks in computer vision, a rich representation can be obtained from images that discriminates different categorizes and therefore obtaining a unsupervised information from images is made possible. However, in the previous works, little attention has been paid to using such unsupervised information for the task of zero-shot learning. In this...
Automatically Learning of Image Features by Using Deep Sparse Networks
, M.Sc. Thesis Sharif University of Technology ; Babaie-Zadeh, Massoud (Supervisor) ; Rabiee, Hamid Reza (Co-Advisor)
Abstract
Data representation plays an important role in machine learning and the performance of machine learning algorithms for instance, in supervised learnings (e.g. classifcation), and unsupervised ones (e.g. image denoising), are heavily influenced by the input applied to them. Regarding the fact that data usually lacks the desirable quality, efforts are always made to make a more desirable representation of data to be used as input to machine learning algorithms. Among many different representation of data, sparse data representation preserves much more information about data while it is simpler than data. We proposed a new stacked sparse autoencoder by imposing power two of smooth L0 norm of...
Designing a Vehicle Counting and Classification System
, M.Sc. Thesis Sharif University of Technology ; Gholampour, Iman (Supervisor)
Abstract
In recent years, Intelligent Transportation Systems (ITS) have received special attentions both in research and in commercial areas. Increased infrastructure facilities, like surveillance cameras, has made this concept even more attainable than before. In this respect, the ability to automatically extract information from traffic images, as one of the key inputs of ITSs, is of great importance. With an increased number of surveillance cameras and the need for more accurate information regarding the road users and their interactions, in order to better city traffic management, building and repairing roads, trip time estimation, number of people per roads estimation and etc, using human...
Image Categorization Using Deep Learning
, M.Sc. Thesis Sharif University of Technology ; Jafari Siavoshani, Mahdi (Supervisor) ; Rabiee, Hamid Reza (Co-Advisor)
Abstract
The representation of data influences the explanation factors of data variations. Thus,the success of learner algorithms depends on the data representation. Our main contribution in this thesis is learning of high level and abstract representation using deep structure. One of the fundamental examples of representation learning is the AutoEncoders. The auto-encoder is a rigid framework that doesn’t consider explanation factors in terms of statistical concepts. So, the auto-encoders can be re-interpreted by seeing the decoder as the statistical model of interest. The role of encoder is a mechanism for inference in the model described by the decoder. Our purpose is to design such model with...
Learning of Alternative Splicing from RNA-seq Data
, M.Sc. Thesis Sharif University of Technology ; Rabiee, Hamid Reza (Supervisor) ; Motahari, Abolfazl (Co-Advisor)
Abstract
We construct and analyse a computational model that predicts the outcome of alternative splicing by recognizing features in RNA sequences. The computational model can be viewed as a “splicing simulator” for a range of healthy human tissues. It takes as input a pre-mRNA sequence surrounding a possibly alternatively spliced exon and estimates the inclusion level of that exon in mature RNA, after splicing occurs. The model is trained using a supervised machine learning framework where the training examples are the alternatively spliced exons, the feature vectors are derived RNA sequences near these exons, and the targets are their corresponding splicing outcomes in healthy individuals. The...
Deep Learning For Recommender Systems
, M.Sc. Thesis Sharif University of Technology ; Soleimani, Mahdieh (Supervisor)
Abstract
Collaborative fltering (CF) is one of the best and widely employed approaches in Recommender systems (RS). This approach tries to fnd some latent features for users and items so it would predict user rates with these features. Early CF methods used matrix factorization to learn users and items latent features. But these methods face cold start as well as sparsity problem. Recent years methods employ side information along with rating matrix to learn users and items latent features. On the other hand, deep learning models show great potential for learning effective representations especially when auxiliary information is sparse. Due to this feature of deep learning, we use deep learning to...
The Application of Deep Learning on Network Traffic Classification
, M.Sc. Thesis Sharif University of Technology ; Jafari Siavoshani, Mahdi (Supervisor)
Abstract
Almost all of the network traffic classification systems use pre-defined extracted features by the experts in computer network. These features include regular expressions, port number, information in the header of different layers and statistical feature of the flow. The main problem of the traffic analysis and anomaly detection system lies in finding appropriate features. The feature extraction is a time consuming process which needs an expert to be done. It is notable that the classification of special kinds of traffic like encrypted traffic is impossible using some subset of mentioned features.The lack of integration in feature detection and classification is also another important issue...
Video Game Testing using Rendering Information and Pattern Recognition
, M.Sc. Thesis Sharif University of Technology ; Habibi, Jafar (Supervisor)
Abstract
After passing many iterations of tests, video games still suffers from major bugs after they’ve been released. One of these problems is the incompatibility between the video game and target audience’s computers. In this thesis, we have proposed a testing strategy by utilizing Cloud Computing infrastructure for video games and Deep Convolutional Neural Networks. We used a simple algorithm to make the graphical output of a video game partially deterministic and thus it’s possible for us to compare different instances of the game in Draw Call level. We’ve used two separate Deep Convolutional Neural Networks for 1. detecting visual artifacts in frames and 2. comparing pairs of Draw Calls for...
Design and Efficient Implementation of ECG-based Detection Algorithm for Dangerous Myocardial Problems
, M.Sc. Thesis Sharif University of Technology ; Hashemi, Matin (Supervisor) ; Vosooghi Vahdat, Bizhan (Co-Advisor)
Abstract
Cardiovascular diseases are the first leading cause of death in the world also in IRAN. Early detection of such problems can decrease the costs also can help to cure the patient but it needs continuous monitoring and automated classification of hearbeats. Mobile devices and wearable gadgets are good solutions which can help patients before visiting the doctor.In this research, an algorithm is introduced which with the help of ECG signal detects dangerous myocardial problems. Our approach is using deep learning method which were not considered much before. In the proposed algorithm ECG signal is processed in order to get features and with dimensionality reduction, input of the network gets...
Design and Efficient Implementation of Deep Learning Algorithm for ECG Classification
, M.Sc. Thesis Sharif University of Technology ; Hashemi, Matin (Supervisor)
Abstract
Cardiovascular diseases are the leading cause of death globally so early diagnosis of them is important. Many researchers focused on this field. First signs of cardiac diseases appear in the electrocardiogram signal. This signal represents the electrical activity of the heart so it’s primarily used for the detection and classification of cardiac arrhythmias. Permanent monitoring of this signal is not possible for specialists so we should do this by means of Artificial Intelligence. In this thesis, we use recurrent neural networks to classify electrocardiogram’s arrhythmias. This deep learning method, use two sources of data to learn from. The first part of data is global for everyone and the...
Implementation of Optical Character Recognition with Deep Learning
, M.Sc. Thesis Sharif University of Technology ; Bagheri Shouraki, Saeed (Supervisor)
Abstract
Optical character recognition (OCR) method has been used in converting printed text into editable text. OCR is very usefuland popular method in various applications. Accuracy of OCRn can be dependent on text preprocessing and segmentation algorithms. Sometimes it is difficult to retrieve text from the image because of different size, style, orientation, complex background of image etc. and Deep learning is a branch of machine learning based on a set of algorithms that attempt to model high-level abstractions in data by using multiple processing layers, with complex structures or otherwise, composed of multiple non-linear transformations and a deep belief network (DBN) is a probabilistic ,...
Performance Improvement of Machine Learning based Intrusion Detection Systems
, M.Sc. Thesis Sharif University of Technology ; Jafari Siavoshani, Mahdi (Supervisor)
Abstract
The rapid growth of computer networks has increased the importance of analytics and traffic analysis tools for these networks, and the increasing importance of these networks has increased the importance of security of these networks and the intrusion detection in these networks. Many studies aimed at providing a powerful way to quickly and accurately detect computer network intrusions, each of which has addressed this issue.The common point of all these methods is their reliance on the features extracted from network traffic by an expert. This strong dependence has prevented these methods from being flexible against new attacks and methods of intrusion or changes in the current normal...
Predicting Patient Clinical Data Using Radiomic Features
, M.Sc. Thesis Sharif University of Technology ; Fatemizadeh, Emadeddin (Supervisor)
Abstract
Genetic differences among patients and cancer types result in different responses to treatments and care from patients. Using personalized medicine, treatments and care can be designed with the specific needs of the patient in mind. To achieve this goal, the informative characteristics of the patient and the disease should be quantified. Quantitative Imaging or Radiomics are concerned with the characterization and quantification of the phenotypical characteristics of the tumors from medical images. Developing handcrafted features is time-consuming and requires the 3D volume of the tumor to be segmented before extracting the features. The segmentation task is considered an open problem and...
Insert Graphical Elements in Multiview Soccer Videos
, M.Sc. Thesis Sharif University of Technology ; Kasaei, Shohreh (Supervisor)
Abstract
In recent decades, many researchers have focused on inferring camera calibration from soccer videos. This task is usually used to provide more information to the audience by adding graphical elements to the field. Indeed, the problem of inserting graphical elements in sport field videos is the problem of calculating projection matrix in continuous frames with which we can insert graphical elements. Basic challenges in this regard are the lack of information in some frames, bad lighting conditions, noise and blur, quick changes of camera viewpoint and radial distortion. Despite previous methods which aimed to propose an algorithm for a specific region of the field, we have introduced a novel...