The design of microstrip coupled resonator filters includes determination the coupling coefficients between filter units. In this paper a novel modeling procedure exploiting prior knowledge neural approach is proposed as an efficient alternative to standard electromagnetic (EM) simulations and models based purely on artificial networks (ANNs). It has similar accuracy EM requires less training d...

In visual semantic navigation, the robot navigates to a target object with egocentric observations and class label of is given. It meaningful task inspiring surge relevant research. However, most existing models are only effective for single-agent single agent has low efficiency poor fault tolerance when conducting more complicated tasks. Multi-agent collaboration can improve strong application...

While widely used in industry, recurrent neural networks (RNNs) are known to have deficiencies dealing with long sequences (e.g. slow inference, vanishing gradients etc.). Recent research has attempted accelerate RNN models by developing mechanisms skip irrelevant words input. Due the lack of labelled data, it remains as a challenge decide which skip, especially for low-resource classification ...