Course Objectives
Students completing this course should be able to construct simple systems using sensors, hierarchical parallel processing, and pattern recognition techniques to perform active perception, clustering, and classification tasks. The course will familiarize students with algorithms and models rooted in probability, statistics, linear algebra, and optimization. Emphasis will be placed on how such models deal with noisy and unlabeled data. At the same time, students will learn how practical considerations of sensor choice and system architecture design can reduce reliance on predictive modeling and improve performance.
Each class will provide students with a mixture of puzzles, models, applications, demonstrations and readings that will help students understand how to implement the process of recognition. Students are expected to use programming languages such as Python, R, MATLAB or C to develop elementary perceptual and recognition systems. Building upon the systems constructed during the course students are expected to develop a final project to be presented at the course's conclusion.
Course Staff
Masatoshi Ishikawa and Carson Reynolds
Keywords
Perception, Statistical Analysis, Sensor Systems
Schedule
- 5/10 Basic Structure of Recognition Systems
- 5/17 Active Perception
- 5/24 Hierarchical Parallel Processing
- 5/31 Information Fusion and Integration
- 6/07 Sensor Models
- 6/14 Feature Extraction
- 6/28 Markovian Pattern Recognition
- 7/12 Particle Filters
- 7/19 Cluster Analysis
- 7/26 Stochastic Optimization
- 8/02 Ensemble Learning
- 8/04 Robotic Recognition
Problem Sets
- Due 6/14 at 23:55 JST Problem Set 1
- Due 7/26 at 23:55 JST Problem Set 2