Here are some selected talks I have given in the past along with links to the corresponding PowerPoint presentations.

The area of automated categorization has grown tremendously over the past decade, due to the increased proliferation of documents in digital form and the need to organize them. An early approach to categorization, knowledge engineering, has largely been superseded by machine learning techniques. In this talk, I will focus on approaches that fall within the machine learning paradigm. I will discuss in detail issues pertaining to three problems within machine learning: document representation, classifier training and classifier evaluation. I will conclude with an exploration of how the web, in particular document link information, can be used to enhance classification techniques. [slides]

An introduction to web services. [slides]

In this talk, I describe MiddleMan, a collection of cooperating proxy servers connected by a local area network (LAN). MiddleMan differs from majority of existing proxy research in that it concentrates exclusively on video. Other approaches are optimized for HTML documents and images. MiddleMan offers several advantages. By caching videos near clients, MiddleMan reduces start-up delays and the possibility of adverse Internet conditions disrupting video playback. Additionally, MiddleMan reduces server load by intercepting a large fraction of server accesses and can easily scale to a large number of users and web video content. It can also be extended to provide other services such as transcoding. [slides]

Despite evidence of rising popularity of video on the web (or VOW), little is known about how users access video. However, such a characterization can greatly benefit the design of multimedia systems such as web video proxies and VOW servers. Hence, this paper presents an analysis of trace data obtained from an ongoing VOW experiment in Luleå University of Technology, Sweden. This experiment is unique as video material is distributed over a high bandwidth network allowing users to make access decisions without the network being a major factor. Our analysis revealed a number of interesting discoveries regarding user VOW access. [slides]

The design of file systems is strongly influenced by measuring the use of existing file systems, such file size distribution and patterns of access. We believe that a similar characterization of video stored on the Internet will help network engineers, codec designers, and other multimedia researchers. We therefore executed an experiment to measure how video data is used on the Web today. I present the results of this experiment and other statistical information characterizing video on the web in this talk [slides]

This talk present techniques for transcoding both in the spatial domain and the frequency domain along with implementation details and current results. In particular, the frequency domain transcoding approach is shown to be faster and more flexible. [slides]

A look into the basics behind sound: acoustics, psycho-acoustics, microphone types, production techniques, audio effects and other factors. [slides]

This talk presents the design of a next generation collaboration system that integrates live and stored video into the WWW framework. The resulting system, which integrates traditional HTML documents with video conferencing, recording, and playback, provides an excellent substrate for distributed group design. [slides]