ㅇ 제목 : 비정상 트래픽/행위 패턴 인식 위한 Dynamic Programming 기술

ㅇ 일시 : 2017년 7월 4일 (화) 16:00~18:00

ㅇ 장소 : ETRI  12연구동 407호

ㅇ 강사 : 신중필교수 (일본 Aizu 대학교)

ㅇ 요약 : 

Dynamic Programming Technique for Recognition of Abnormal Traffic / Behavior Patterns

Nowadays various kinds of algorithm is introduced for pattern recognition. Dynamic programming (DP) algorithm is one of powerful techniques for pattern recognition. The main advantages of the resulting DP algorithms are:

(1) The problem can be solved without knowing the statistical properties of the measured signal in detail.

(2) A high degree of stability against local disturbances is achieved.

(3) The algorithm solution after a finite number of computation steps.

(4) The algorithm allows a straightforward machine implementation.

The algorithm derived for the optimal path problem is applicable to a wide variety of the pattern recognition problems and their solutions by dynamic programming. The concept of dynamic programming is especially useful in solving combinatorial optimization problems which can be broken down into a sequence of optimization stages and in which there are only a small number of possible choices or decision to be taken at each optimization stage. DP Technique can be applied for the recognition of abnormal traffic / behavior patterns. In this seminar, we explain basic DP and advanced algorithm with improved performance for the recognition of abnormal traffic / behavior patterns.