Human motion similarity is practiced in many fields, including action recognition, anomaly detection, and human performance evaluation. While many computer vision tasks have benefited from deep learning, measuring motion similarity has attracted less attention, particularly due to the lack of large datasets. To address this problem, we introduce two datasets: a synthetic motion dataset for model training and a dataset containing human annotations of real-world video clip pairs for motion similarity evaluation. Furthermore, in order to compute the motion similarity from these datasets, we propose a deep learning model that produces motion embeddings suitable for measuring the similarity between different motions of each human body part. The network is trained with the proposed motion variation loss to robustly distinguish even subtly different motions. The proposed approach outperforms the other baselines considered in terms of correlations between motion similarity predictions and human annotations while being suitable for real-time action analysis. Both datasets and codes are released to the public.

The task of sentence completion, which aims to infer the missing text of a given sentence, was carried out to assess the reading comprehension level of machines as well as humans. In this work, we conducted a comprehensive study of various approaches for the sentence completion based on neural language models, which have been advanced in recent years. First, we revisited the recurrent neural network language model (RNN LM), achieving highly competitive results with an appropriate network structure and hyper-parameters. This paper presents a bidirectional version of RNN LM, which surpassed the previous best results on Microsoft Research (MSR) Sentence Completion Challenge and the Scholastic Aptitude Test (SAT) sentence completion questions. In parallel with directly applying RNN LM to sentence completion, we also employed a supervised learning framework that fine-tunes a large pre-trained transformer-based LM with a few sentence-completion examples. By fine-tuning a pre-trained BERT model, this work established state-of-the-art results on the MSR and SAT sets. Furthermore, we performed similar experimentation on newly collected cloze-style questions in the Korean language. The experimental results reveal that simply applying the multilingual BERT models for the Korean dataset was not satisfactory, which leaves room for further research.

Voice conversion is a task of synthesizing an utterance with target speaker's voice while maintaining linguistic information of the source utterance. While a speaker can produce varying utterances from a single script with different intonations, conventional voice conversion models were limited to producing only one result per source input. To overcome this limitation, we propose a novel approach for voice conversion with diverse intonations using conditional variational autoencoder (CVAE). Experiments have shown that the speaker's style feature can be mapped into a latent space with Gaussian distribution. We have also been able to convert voices with more diverse intonation by making the posterior of the latent space more complex with inverse autoregressive flow (IAF). As a result, the converted voice not only has a diversity of intonations, but also has better sound quality than the model without CVAE.