Yin Tong

Hao Lan, Yin Tong, Carla Seatzu et al.

Detectability describes the property of a system whose current and the subsequent states can be uniquely determined after a finite number of observations. In this paper, we developed a novel approach to verifying strong detectability and periodically strong detectability of bounded labeled Petri nets. Our approach is based on the analysis of the basis reachability graph of a special Petri net, called Verifier Net, that is built from the Petri net model of the given system. Without computing the whole reachability space and without enumerating all the markings, the proposed approaches are more efficient.

Mincong Luo, Yin Tong, Jiachi Liu

One less addressed issue of deep reinforcement learning is the lack of generalization capability based on new state and new target, for complex tasks, it is necessary to give the correct strategy and evaluate all possible actions for current state. Fortunately, deep reinforcement learning has enabled enormous progress in both subproblems: giving the correct strategy and evaluating all actions based on the state. In this paper we present an approach called orthogonal policy gradient descent(OPGD) that can make agent learn the policy gradient based on the current state and the actions set, by which the agent can learn a policy network with generalization capability. we evaluate the proposed method on the 3D autonomous driving enviroment TORCS compared with the baseline model, detailed analyses of experimental results and proofs are also given.