Perceptive Computing

Instructor: Yang Cai

12 units

05-499 D and 05-899 D

Perceptive Computing is a new field that combines visual cognition,

machine vision and visualization. It is a computational simulation of

human insight or intuition in massive data analysis, problem solving and

learning. Human perceptions can capture complex patterns, relationships

and exceptions in a data set. Perceptive computing is a way to summarize

seemingly disjoint data into significant parts and pass the summary

information to decision entities. However, many "intuitive" algorithms

are not scalable. In many cases, we have to compromise factors such as,

accuracy, scale and speed, or use radical approaches such as dimensional

compression, randomization, etc.

This is an advanced project-oriented course. The goal of the class is to

develop novel vision algorithms for solving real world problems. For

example, NASA satellite data processing, in-vitro cell motion

classification, tongue-based inspection, behavior measurement from

head-mounted video or surveillance video systems, etc. Actual field data

will be provided to the class. Each project will have four progressive

releases in one semseter: concept, skeleton, alpha and final release. The

final product will be a paper and a demo.

Requirements:

matrix computing and a programming language: C or Java.

Need the pre-approval from the instructor.