The data of social media has received much attention to observe and predict real-world events. For example, It is used to predict financial markets, products demand, and voter turnout. While these works regards social media as a sensor of real world, as social media become more popular, it become more natural to think social media significantly effects on real worlds events. The canonical example might be cryptocurrencies, where supply and demand are more susceptible to investor sentiment and therefore interactions within social media cause significant effects on the price of them. On the hypothesis that social media actuate real-world events, we propose a neural network based model to predict the price fluctuations of financial assets, including cryptocurrencies. We model the effect of social media which cannot be directly observed, using an end-to-end neural network, Recurrent Neural Network. By simulating the effect within the social media, we show that the method that models the effect of social media on financial markets can observe and predict the price fluctuations of cryptocurrencies more precisely and stably. By analyzing the model, we suggest that networks within social media can be influential relationships throughout time, even if they are not directly connected, and that the intensity of the influence from social media on financial markets varies depending on the nature of the financial assets.
Pointer-generator, which is the one of the strong baselines in neural summarization models, generates summaries by selecting words from a set of words (output vocabulary) and words in source documents. A conventional method for constructing output vocabulary collects highly frequent words in summaries of training data. However, highly frequent words in summaries could be usually a high possibility to be frequent in source documents. Thus, an output vocabulary constructed by the conventional method is redundant for pointer-generator because pointergenerator can copy words in source documents. We propose a vocabulary construction method that selects words included in each summary but not included in its source text of each pair. Experimental results on CNN/Daily Mail corpus and NEWSROOM corpus showed that our method contributes to improved ROUGE scores while obtaining high ratios of generating novel words that do not occur in source documents.
Driver pose estimation is a key component in driver monitoring systems, which is helpful for driver anomaly detection. Compared with traditional human pose estimation, driver pose estimation is required to be fast and compact for embedded systems. We propose fast and compact driver pose estimation that is composed of ShuffleNet V2 and integral regression. ShuffleNet V2 can reduce computational expense, and integral regression reduce quantization error of heat maps. If a driver suddenly gets seriously ill, the head of the driver is out of view. Therefore, in addition to localizing body parts, classifying whether each body part is out of view is also crucial for driver anomaly detection. We also propose a novel model which can localize and detect each body part of the driver at once. Extensive experiments have been conducted on a driver pose estimation dataset recorded with near infrared camera which can capture a driver at night. Our method achieves large improvement compared to the state-of-the-art human pose estimation methods with limited computation resources. Futhermore, We perform an ablation study of our method which composed of ShuffleNet V2, integral regression, and driver body parts detection. Finally, we show experimental results of each driver action for driver monitoring systems.
The purpose of this paper is to report on the design and development of a Venn diagram and Yes/No chart learning system for fostering computational thinking and its evaluation. We developed a Yes/No chart learning system as a web application. This system was developed using HTML5 and vis.js. To evaluate this system’s effectiveness, 20 university students and 4 high school students were given an experiential learning exercise. As a result, 18 students were able to use this system to explain the classification better. The median computational thinking score after using the system was significantly higher than the median computational thinking score before using the system. It was confirmed that this system is useful in fostering computational thinking through experiential learning.