Abstract

We propose a model of coverage patterns (CPs) and approaches for extracting CPs from transactional databases. The model is motivated by the problem of banner advertisement placement in e-commerce web sites. Normally, an advertiser expects that the banner advertisement should be displayed to a certain percentage of web site visitors. On the other hand, to generate more revenue for a given web site, the publisher makes efforts to meet the coverage demands of multiple advertisers. Informally, a CP is a set of non overlapping items covered by certain percentage of transactions in a transactional database.The CPs do not satisfy the downward closure property. Efforts are being made in the literature to extract CPs using level-wise pruning approach. In this paper, we propose CP extraction approaches based on pattern growth techniques. Experimental results show that the proposed pattern growth approaches improve the performance over the level-wise pruning approach. The results also show that CPs could be used in meeting the demands of multiple advertisers.

Abstract

Online advertising provides an opportunity for product sellers and service providers to reach customers and has become a key factor in the growth of economy. It is a major source of revenue for the major search engine and social networking sites. Search engine, context-specific and banner advertising are the major modes of online advertising. The banner advertisement mode has certain advantages over other modes of advertising. Currently, the number of websites registered comes to a billion. Each day, a typical website receives the number of visitors ranging from hundreds to millions. In a few years, the entire population of the globe is going to be connected to Internet and browse websites. It is possible for a product seller or service provider to reach every potential customer through banner advertising. In this paper, a framework is proposed to harvest the pages views of web by forming the clusters of similar websites. Rather than managing a single website, the publisher manages the aggregated advertising space of a collection of websites. As a result, the advertisement space could be expanded significantly and it will provide the opportunity for increased number of publishers to market the aggregated advertisement space of millions of websites to advertisers for reaching potential customers. It will also help in balancing the management of banner advertising market.

Abstract

Crowd management is a complex, challenging and crucial task. Lack of appropriate management of crowd has, in past, led to many unfortunate stampedes with significant loss of life. To increase the crowd management efficiency, we deploy automated real time detection of stampede prone areas. Then, we use robotic agents to aid the crowd management police in controlling the crowd in these stampede prone areas. While doing so, we aim for minimum interference by robotic agents in our environment. Thereby not disturbing the ambiance and aesthetics of the place. We evaluate the effectiveness of our model in dealing with difficult scenarios like emergency evacuation and presence of localized congestion. Lastly, we simulate a multi agent system based on our model and use it to illustrate the utility of robotic agents for detecting and reducing congestion.

Abstract

Preventing traffic congestion by forecasting near time traffic flows is an important problem as it leads to effective use of transport resources. Social network provides information about activities of humans and social events. Thus, with the help of social network, we can extract which humans will attend a particular event (in near time) and can estimate flow of traffic based on it. This opens up a wide area of research which poses need to have a framework for traffic management that can capture essential parameters of real-life behaviour and provide a way to iterate upon and evaluate new ideas. In this paper, we present building blocks of a framework and a system to simulate a city with its transport system, humans and their social network. We emphasize on relevant parameters selected and modular design of the framework. Our framework defines metrics to evaluate congestion avoidance strategies. To show utility of the framework, we present experimental studies of few strategies on a public transport system.

Abstract

A queue is required when a service provider is not able to handle jobs arriving over the time. In a highly flexible and dynamic environment, some jobs might demand for faster execution at run-time especially when the resources are limited and the jobs are competing for acquiring resources. A user might demand for speed up (reduced wait time) for some of the jobs present in the queue at run time. In such cases, it is required to accelerate (directly sending the job to the server) urgent jobs (requesting for speed up) ahead of other jobs present in the queue for an earlier completion of urgent jobs. Under the assumption of no additional resources, such acceleration of jobs would result in slowing down of other jobs present in the queue. In this paper, we formulate the problem of Speed Up Scheduling without acquiring any additional resources for the scheduling of on-line speed up requests posed by a user at run-time and present algorithms for the same. We apply the idea of Speed Up Scheduling to two different domains-Web Scheduling and CPU Scheduling. We demonstrate our results with a simulation based model using trace driven workload and synthetic datasets to show the usefulness of Speed Up scheduling. Speed Up provides a new way of addressing urgent jobs, provides a different evaluation criteria for comparing scheduling algorithms and has practical applications.

Abstract

Preventing traffic congestion by forecasting near time traffic flows is an important problem as it leads to effective use of transport resources. Social network provides information about activities of humans and social events. Thus, with the help of social network, we can extract which humans will attend a particular event (in near time) and can estimate flow of traffic based on it. This opens research area to build a framework for traffic management that can capture essential parameters of real-life behavior and provide a way to iterate upon and evaluate new ideas. In this paper, we present building blocks of such framework and a system to simulate a city with its transport system, humans and their social network. We emphasize on relevant parameters selected and modular design of the framework. To show the utility of the framework, we present experimental studies of few strategies on a public transport system.

Abstract

Crowd management is a complex, challenging and crucial task. Lack of appropriate management of crowd has, in past, led to many unfortunate stampedes with significant loss of life. To increase the crowd management efficiency, we deploy automated real time detection of stampede prone areas. Further, we use robotic agents for aiding the crowd management police in controlling the crowd in these stampede prone areas. Lastly, we simulate a multi agent system based on our model and use it to illustrate the utility and viability of robotic agents for detecting and reducing congestion.

Abstract

We present an automated disease term classification model using machine learning techniques that classifies a medical term to a specific disease class. We work on five particular diseases: Cancer, AIDS, Arthritis, Diabetes and heart related ailments. We identify and classify medical terms like drug names, symptoms, abbreviations, disease names, tests, etc., into their specific diseases classes. The results illustrate that our model for disease term classification finds all disease term classes with an average F-score of 0.966.

Abstract

Sonorant segmentation of speech signals is critical in developing Automatic Speech Recognition (ASR) systems, audio search systems and for automatic segmentation of speech corpora. In this work, acoustic features based on excitation source and vocal tract system characteristics of sonorant sounds are proposed for segmentation of sonorant regions in continuous speech. The features are based on zero frequency resonator signal energy, strength of excitation and dominant resonance frequency around epochs. An algorithm is developed to relate these features in hierarchical manner using knowledge-based approach. The performance of the proposed algorithm is studied on three different datasets, at varying levels of degradation. TIMIT database is used to test the validity and AMI meeting corpus and Telugu (an Indian language) dataset are considered to test the utility of the proposed features.

Abstract

In this paper, we investigate two different recurrent neural network (RNN) architectures: Elman RNN and recently proposed clockwork RNN [1] for statistical parametric speech synthesis (SPSS). Of late, deep neural networks are being used for SPSS which involve predicting every frame independent of the previous predictions, and hence requires post-processing for ensuring smooth evolution of speech parameters. RNNs, on the other hand, are intuitively better suited for the task as they inherently model temporal dependencies, but were restricted in use because of the difficulty in training. Lately, techniques such as sparse initialization, Nesterov’s accelerated gradient, gradient clipping and leaky integration (LI) have been shown to overcome this difficulty. We study the utility of these techniques for SPSS task. In addition, we show that clockwork RNN is equivalent to an Elman RNN with a particular form of LI. This perspective enables us to understand the reason why a simple Elman RNN with LI units performs well on sequential tasks.