Neural question generation (NQG) is the task of generating a question from a given passage with deep neural networks. Previous NQG models suffer from a problem that a significant proportion of the generated questions include words in the question target, resulting in the generation of unintended questions. In this paper, we propose answer-separated seq2seq, which better utilizes the information from both the passage and the target answer. By replacing the target answer in the original passage with a special token, our model learns to identify which interrogative word should be used. We also propose a new module termed keyword-net, which helps the model better capture the key information in the target answer and generate an appropriate question. Experimental results demonstrate that our answer separation method significantly reduces the number of improper questions which include answers. Consequently, our model significantly outperforms previous state-of-the-art NQG models.

Conventional full search block matching algorithm (BMA) for motion estimation is computationally intensive and the resulting hardware cost is very high. In this paper, we present an efficient BMA using a 16:1 subsampling technique, and describe its hardware design. The algorithm reduces the number of pixels in calculating the mean absolute difference at each search location, instead of reducing the search locations. The algorithm is an extension of the BMA with 4:1 subsampling proposed by Lin and Zaccarin (1993) such that the amount of computation is reduced by a factor of 4 (16 compared to the full search BMA) while producing similar performance. The algorithm can efficiently be designed into a hardware for real-time applications.