Although game theory is well-established for linear games, where there is no repetition, and games with loops are often disregarded, due to it’s nature of loops being hard to handle, as a player can stay in a loop forever to avoid losing, and thus it is hard to determine whether a the player has a winning strategy efficiently. Due to this lack of exploration of games with loops, modern board games usually have this quality of being complex in terms of the variety of moves, and having some loops. This research proposes an intuitive approach to extend the Sprague-Grundy Theorem from linear games to games that contain cycles to find a computational algorithm to efficiently calculate the wining stragegy.

With the most professional leagues, professional players, and national competitions, soccer is perhaps the most popular sport in the world. One of the reasons why soccer is so popular is because it is simple: two teams trying to score in each other’s goal only using their feets and heads. However, when we go deeper inside, there are numerous factors that can subtly or significantly impact the entire result of the game. The factor that this paper focuses on is the ball distribution between each player. It is common sense that teams with equal contribution from every player indicate better teamwork, and thus they are more likely to be stronger than teams that rely on one or two key players. In order to qualify this idea and devise useful strategies accordingly, we employed Markov chain and Principal Component Analysis (PCA). Through the Markov chain, we modeled a soccer game into a system (team) of eleven sections (players) continuously transitioning (giving passes) to other sections until they score a goal. Through PCA, we compared the patterns we found from the Markov chain modeling to other soccer statistics to evaluate how related the pattern we found is to the victory of the soccer team compared to other well-known soccer statistics. The evaluation of our approach shows promising results in analyzing soccer and constructing the most ideal player formation to win the game.

From hair implants to plastic surgeries that reduce wrinkles, there is numerous evidence that reflect the full-grown adult’s interest in appearances. The most classic, but also the most difficult to answer, question in this field is "what makes a face attractive?" Indeed, there are certain facial features that are commonly seen in attractive faces, but it is hard to generalize them since the beauty standard differs from person to person and sometimes, it is not one or two features, but the harmony of every feature that makes the face look more appealing. In order to find a more objective and reliable way to assess attractiveness of different faces, we employed a face mesh technique based on a lightweight statistical analysis method called Procrustes Analysis and Principal Component Analysis (PCA). As a result, we constructed an algorithm that predicts the appearance rating of a random face. This system can be exploited in fields like dating apps, plastic surgery counseling, or cosmetic developments.

Virtual learning platforms have extended the learning experience beyond the constraints of space and time. Learning new language skills in virtual environments is becoming more and more convenient and efficient. In this work, learning English as a second language (ESL) is tested. Initial results showed that it is not only efficient to learn ESL with this tool, but it also helped users to develop good interaction skills with the computer-generated 3D world without any fear or shyness. Current virtual learning platforms suffers from distraction inherited by designing learning activities that requires user interaction, such as pressing the keyboard and clicking the mouse to control the avatar and background. In our research work, we employed a new software implementation stack that include a unity virtual environment, an applied Speech Recognition (SR) capability, and a Natural Language Processing (NLP) component to overcome these distractions and improve the Human-Computer Interaction (HCI) capability without the need for a keyboard and mouse for the dynamic avatar’s movement and backgrounds.

The overload of online data creates a problem in filtering information that is appropriate to a user, which is why many companies and websites utilize recommendation systems. Finding products and contents that match users' interests has become crucial for online content-providers, and users are always in search for content that suit their preferences. It is becoming more and more difficult for users to find the content they are looking for with the rate of content increasing in the world, thus the importance of recommender systems has increased more than ever. Yet, recommendation systems have not been perfected and shows much room for research and development. This paper therefore presents a new hybrid method, integrating a parallel collaborative and content-based filtering system with a ranking system based on a user profile with the addition of a time variable to represent the time sensitive characteristic of content and interest. The evaluation of our approach shows promising results in the cinematographic field and indicates further potential for development.

The importance of medical facilities to citizens is increasing due to frequently occurring pandemics and increased interest in primary health. Although these facilities must be spread widely throughout places to be easily accessed to everyone, it is typically concentrated in central, urban areas. The quality of medical service can be evaluated by the ratio of hospitals of different sizes--clinic, hospital, and territory hospital--since large hospitals offer a higher quality medical care. This research aims to analyze the correlation between each district's median income level in Seoul and medical service quality using a quantitative analysis method based on correlation coefficient analysis. The correlation analysis showed that territory hospitals have a strong correlation with the district's income level, while general hospitals and clinics have a higher correlation with the number of residents and workers.

In the modern-day stock markets, accuracy and timeliness of price predictive methods can be key in making a profit and edging out the competition. Stockbrokers utilize predictive systems based on these methods that rely on a plethora of data to assist in decision making. To work towards developing our own system that will eventually utilize big data, we performed research with recurrent neural networks (RNNs) and support vector machines (SVMs) to derive a backing method for the predictive system. In this paper we discuss the experimentation we have performed, and the proposed approach derived from the results of the experimentation.

The Internet of Mobile Things encompasses stream data being generated by sensors, network communications that pull and push these data streams, as well as running processing and analytics that can effectively leverage actionable information for transportation planning, management, and business advantage. Edge computing emerges as a new paradigm that decentralizes the communication, computation, control and storage resources from the cloud to the edge of the network. This paper proposes an edge computing platform where mobile edge nodes are physical devices deployed on a transit bus where descriptive analytics is used to uncover meaningful patterns from real-time transit data streams. An application experiment is used to evaluate the advantages and disadvantages of our proposed platform to support descriptive analytics at a mobile edge node and generate actionable information to transit managers.