Two-Stream Retentive Long Short-Term Memory Network for Dense Action Anticipation.

Analyzing and understanding human actions in long-range videos has promising applications, such as video surveillance, automatic driving, and efficient human-computer interaction. Most researches focus on short-range videos that predict a single action in an ongoing video or forecast an action several seconds earlier before it occurs. In this work, a novel method is proposed to forecast a series of actions and their durations after observing a partial video. This method extracts features from both frame sequences and label sequences. A retentive memory module is introduced to richly extract features at salient time steps and pivotal channels. Extensive experiments are conducted on the Breakfast data set and 50 Salads data set. Compared to the state-of-the-art methods, the method achieves comparable performance in most cases.

Application of multi-objective controller to optimal tuning of PID gains for a hydraulic turbine regulating system using adaptive grid particle swam optimization.

A hydraulic turbine regulating system (HTRS) is one of the most important components of hydropower plant, which plays a key role in maintaining safety, stability and economical operation of hydro-electrical installations. At present, the conventional PID controller is widely applied in the HTRS system for its practicability and robustness, and the primary problem with respect to this control law is how to optimally tune the parameters, i.e. the determination of PID controller gains for satisfactory performance. In this paper, a kind of multi-objective evolutionary algorithms, named adaptive grid particle swarm optimization (AGPSO) is applied to solve the PID gains tuning problem of the HTRS system. This newly AGPSO optimized method, which differs from a traditional one-single objective optimization method, is designed to take care of settling time and overshoot level simultaneously, in which a set of non-inferior alternatives solutions (i.e. Pareto solution) is generated. Furthermore, a fuzzy-based membership value assignment method is employed to choose the best compromise solution from the obtained Pareto set. An illustrative example associated with the best compromise solution for parameter tuning of the nonlinear HTRS system is introduced to verify the feasibility and the effectiveness of the proposed AGPSO-based optimization approach, as compared with two another prominent multi-objective algorithms, i.e. Non-dominated Sorting Genetic Algorithm II (NSGAII) and Strength Pareto Evolutionary Algorithm II (SPEAII), for the quality and diversity of obtained Pareto solutions set. Consequently, simulation results show that this AGPSO optimized approach outperforms than compared methods with higher efficiency and better quality no matter whether the HTRS system works under unload or load conditions.