Sharif Digital Repository

Web crawlers are one of the most crucial components in search engines and their optimization would have a great effect on improving the searching efficiency. In this paper, we introduce an intelligent crawler called Gcrawler that uses a genetic algorithm for improving its crawling performance. Gcrawler estimates the best path for crawling on one hand and expands its initial keywords by using a genetic algorithm during the crawling on the other hand. This is the first crawler that acts intelligently without any relevance feedback or training. All the processes are online and there is no need for direct interaction with the users. © 2005 IEEE  

Artificial Life can be used as an agent training approach in large state spaces. This paper presents an artificial life method to increase the training speed of some speech recognizer agents which where previously trained by genetic algorithms. Using this approach, vertical training (genetic mutations and selection) is combined with horizontal training (individual learning through reinforcement learning) and results in a much faster evolution than simple genetic algorithm. The approach is tested and a comparison with GA cases on a standard speech data base is presented. COPYRIGHT © ENFORMATIKA  

Grammar induction, also known as grammar inference, is one of the most important research areas in the domain of natural language processing. Availability of large corpora has encouraged many researchers to use statistical methods for grammar induction. This problem can be divided into three different categories of supervised, semi-supervised, and unsupervised, based on type of the required data set for the training phase. Most current inductive methods are supervised, which need a bracketed data set for their training phase; but the lack of this kind of data set in many languages, encouraged us to focus on unsupervised approaches. Here, we introduce a novel approach, which we call... 

Statistical bias correction methods are inferred relationships between inputs and outputs. The constructed functions are based on available observations, which are limited in time and space. This study investigates the ability of regression models (linear and nonlinear) to regionalize a domain by defining a minimum number of training pixels necessary to achieve a good level of bias correction performance. Linear regression is used to divide northern South America into five regions. To correct the biases of temperature and precipitation, an artificial neural network (ANN) model was trained with selected pixels within each region and then used to reproduce bias-corrected temperature and... 

Multi-label classification has received considerable interest in recent years. Multi-label classifiers usually need to address many issues including: handling large-scale datasets with many instances and a large set of labels, compensating missing label assignments in the training set, considering correlations between labels, as well as exploiting unlabeled data to improve prediction performance. To tackle datasets with a large set of labels, embedding-based methods represent the label assignments in a low-dimensional space. Many state-of-the-art embedding-based methods use a linear dimensionality reduction to map the label assignments to a low-dimensional space. However, by doing so, these... 

We present and evaluate Deep Private-Feature Extractor (DPFE), a deep model which is trained and evaluated based on information theoretic constraints. Using the selective exchange of information between a user's device and a service provider, DPFE enables the user to prevent certain sensitive information from being shared with a service provider, while allowing them to extract approved information using their model. We introduce and utilize the log-rank privacy, a novel measure to assess the effectiveness of DPFE in removing sensitive information and compare different models based on their accuracy-privacy trade-off. We then implement and evaluate the performance of DPFE on smartphones to... 

Web-based E-Leaning Systems (WELSs) have emerged as new means of skill training and knowledge acquisition, encouraging both academia and industry to invest resources in the adoption of these systems. Traditionally, most pre- and post-adoption tasks related to evaluation are carried out from the viewpoints of technology. Since users have been widely recognized as being a key group of stakeholders in influencing the adoption of information systems, their attitudes about these systems are considered as pivotal. Therefore, based on the theory of multi-criteria decision making and the research results concerning user satisfaction in the fields of human-computer interaction and information... 

Social robots may soon be able to play an important role in expanding communication with the deaf. Based on the literature, adaptive user interfaces lead to greater user acceptance and increased teaching efficiency compared to non-adaptive ones. In this paper, we build a robotic architecture able to simultaneously adjust a robot’s teaching parameters according to both the user’s past and present performance, adapt the content of the training, and then implement it on the RASA robot to teach sign language based on these parameters in a manner similar to a human teacher. To do this, a word to teach in sign language, repetition, speed, and emotional valence were chosen to be adaptive using a... 

Deep learning architectures employ various regularization terms to handle different types of priors. Non-smooth regularization terms have shown promising performance in the deep learning architectures and a learning framework has recently been proposed to train autoencoders with such regularization terms. While this framework efficiently manages the non-smooth term during training through proximal operators, it is limited to autoencoders and suffers from low convergence speed due to several optimization sub-problems that must be solved in a row. In this paper, we address these issues by extending the framework to general feed-forward neural networks and introducing variable extrapolation... 

The objective of the present study is to depict an inclusive Construction Waste Management (CWM) plan looking at the total project life cycle. This holistic approach is called Integrated Construction Waste Management (ICWM). This research program has been conducted through several consecutive academic dissertations at Civil Engineering Department of SUT and was firstly aimed to identify waste sources throughout project life cycle. Concurrent research efforts were focused on project delivery methods evaluation, e.g. contract type effect on waste generation amount along with appropriate guidelines/incentives development that could promote ICMW. These studies were conducted via field... 

Multi-label classification has received considerable interest in recent years. Multi-label classifiers usually need to address many issues including: handling large-scale datasets with many instances and a large set of labels, compensating missing label assignments in the training set, considering correlations between labels, as well as exploiting unlabeled data to improve prediction performance. To tackle datasets with a large set of labels, embedding-based methods represent the label assignments in a low dimensional space. Many state-of-the-art embedding-based methods use a linear dimensionality reduction to map the label assignments to a low-dimensional space. However, by doing so, these... 

We present and evaluate Deep Private-Feature Extractor (DPFE), a deep model which is trained and evaluated based on information theoretic constraints. Using the selective exchange of information between a user's device and a service provider, DPFE enables the user to prevent certain sensitive information from being shared with a service provider, while allowing them to extract approved information using their model. We introduce and utilize the log-rank privacy, a novel measure to assess the effectiveness of DPFE in removing sensitive information and compare different models based on their accuracy-privacy trade-off. We then implement and evaluate the performance of DPFE on smartphones to... 

This paper is focused on video event recognition based on frame level CNN descriptors. Using transfer learning, the image trained descriptors are applied to the video domain to make event recognition feasible in scenarios with limited computational resources. After fine-tuning of the existing Convolutional Neural Network (CNN) concept score extractors, pre-trained on ImageNet, the output descriptors of the different fully connected layers are employed as frame descriptors. The resulting descriptors are hierarchically post-processed and combined with novel and efficient pooling and normalization methods. As major contributions of this work to the video event recognition, we present a... 

This paper is focused on video event recognition based on frame level convolutional neural network (CNN) descriptors. Using transfer learning, the image trained descriptors are applied to the video domain to make event recognition feasible in scenarios with limited computational resources. After fine-tuning of the existing CNN concept score extractors, pretrained on ImageNet, the output descriptors of the different fully connected layers are employed as frame descriptors. The resulting descriptors are hierarchically postprocessed and combined with novel and efficient pooling and normalization methods. As major contributions of this paper to the video event recognition, we present a... 

This paper is focused on video event recognition based on frame level convolutional neural network (CNN) descriptors. Using transfer learning, the image trained descriptors are applied to the video domain to make event recognition feasible in scenarios with limited computational resources. After fine-tuning of the existing CNN concept score extractors, pretrained on ImageNet, the output descriptors of the different fully connected layers are employed as frame descriptors. The resulting descriptors are hierarchically postprocessed and combined with novel and efficient pooling and normalization methods. As major contributions of this paper to the video event recognition, we present a...