HPI Kolloquien: "Multimedia Analysis" und "Visualization Techniques"

"Multimedia Analysis with Deep Learning" / Dr. Haojin Yang (HPI)


From the year 2006 “Deep Learning” (DL) has attracted more and more attentions in both academia and industry. Recently DL gives us break-record results in many novel areas, as e.g., beating human in strategic game systems like Go (Google’s AlphaGo), self-driving cars, achieving dermatologist-level classification of skin cancer etc. The widely accepted success factors of Deep Learning include the rejuvenation of stacked artificial neural networks (getting deeper and wider), the appearance of large labeled datasets (over millions of training samples), and the evolution of computation power (applying GPU acceleration and distributed computing). Multimedia data is one of most suitable objectives for deep learning research, because of its multiple modalities. Multimedia consists of visual, textual and auditory content. This specific feature could enable DL algorithms to learn fused representations from hybrid resources, which illustrate common semantic meaning. On the other hand, DL is a data-driven technology which makes it highly suitable to process massive amounts of multimedia data. This talk will present several current research topics of multimedia analysis based on DL technologies.

More about Haojin Yang

Host: Prof. Christoph Meinel

"Visualization Techniques for Massive Data Sets" / Dr. Matthias Trapp (HPI)


Interactive visualization techniques are a fundamental prerequisite for visual computing and visual analysis tasks. During recent years, due to increase of massive parallel computing power at hands of consumer graphics hardware, various approaches has been researched to increase both rendering performance and visual quality in computer graphics and interactive visualization. Most of these approaches rely on preprocessed input data to handle the ubiquitous increasing data volume. This talk presents an approach for implementing stages of the visualization pipeline (especially filtering, mapping, and rendering) using GPU capabilities in order to decrease the amount of preprocessing required for the visualization of massive data sets. Based on a brief introduction to visualization, it demonstrates the application of this approach to visualization techniques in the domains of geovisualization and software visualization.

More about Dr. Matthias Trapp

Host: Prof. Jürgen Döllner