9661

8905

كارشناسي ارشد

كامپيوتر - هوش مصنوعي

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

1393

[ده]،68ص.: مصور،جدول،نمودار

ص.ع.به فارسي و انگليسي

عبدالرضا ميرزائي

93/12/20

رسول امير فتاحي، مهران صفاياني

مهندسي برق و كامپيوتر

ID8905

Human Action Recognition Using Probabilistic Graphical Models Amir Hossein Saleki a saleki@ec iut ac ir September 2014 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language FarsiSupervisor Abdolreza Mirzaei mirzaei@cc iut ac irAbstractProbabilistic Graphical Models is a powerful framework for representation and inference in probabilitydistributions with many random variables Sequence labeling is one of the most challenging problems in thefield of pattern recognition In this problem we want to predict an output vector given a sequence ofobservations It has many applications such as part of speech tagging handwritten text recognition speechrecognition protein secondary structure prediction and human action recognition So far a lot ofprobabilistic models such as HMMs MEMMs CRFs HCRF and CNFs have been used to solve thisproblem efficiently and accurately In models like CRF the distribution over output variables is a log linearfunction of observations In practice the relation between inputs and outputs is highly non linear In thisthesis we propose a model to assign a single label to a sequence of observations which capture the non linearity between inputs and outputs by using a layer of ANFIS networks We evaluate the proposed modelon the task of human action recognition using skeleton data and show that our model achieves better resultsthan models like HCRF and CNF Key WordsProbabilistic Graphical Models Conditional Random Fields Hidden ConditionalRandom Fields Conditional Neural Fields Human Action Recognition

عبدالرضا ميرزائي

رسول امير فتاحي، مهران صفاياني