In recent years, the spread and abuse of drugs has always been the focus of social attention. The spread of drugs can cause serious problems in our society and endanger the lives, property and health of the public. Although there have been many efforts in the domain of ant i drug, the number of guilty persons in drug cases is still rising. Therefore, there is a need to find a way to promote anti drug awareness. More applications can achieved with an artificial intelligence computer, such as chatbot. There are several advantages to using a chatbot: it can work around the clock, effectively reducing labor costs, can be deployed on social media mess aging platforms and easy to use. Therefore we propose Anti Drug Friend that utilize chatbot to advocate general knowledge ab out drugs and deepening anti drug awareness for people. Currently, our Anti Drug Friend is already serving on LINE. For a n anti drug domain, our chatbot can answer user: Information about drugs, How to deal with drug problems, Pushing the latest news and information about drugs.

In recent years, the application field of drones has rapidly expanded, and they are attracting attention as a means of gathering information in disaster areas and as a means of disaster relief. Currently, drones are mainly controlled manually or by programmed control with pre-defined actions. Manual control limits the flight range to the transmitter's communication range and requires constant monitoring of the drone in flight, making it difficult to operate the drone when access to the disaster area is difficult or visibility is poor. In addition, since programmed control sets the drone's behavior in advance, stable control in response to changes in the flight environment due to a disaster may be difficult in emergency surveys during a disaster. The purpose of this study is to verify the applicability of model-based control, which is based on a mathematical model of the environment, and model-free control, which obtains optimal control rules from interaction with the environment, to drone’s controller for realizing autonomous drone flight in environments that are difficult to operate with existing control methods. Specifically, we compared the performance of model predictive control as model-based control and reinforcement learning as model-free control by computer experiments, and verified that the characteristics of both control system differ depending on the nature of the disturbance.

In this paper, the process of long-term development of the MiLog platform, which is a software prototyping platform for agent-based, multiagent-based, and mobile agent-based systems, first officially published in 2000, is explained. A set of analyses on how the platform has been developed, extended, and maintained to fit to the current computing platform to support prototype systems reproducible, are explained.

Systems with multiple moving agents work inherently in a distributed environment with many uncertainties for timing assumptions such as delays. Thus, perfect executions of multi-agent path planning, which assume synchronized action between agents, become more challenging with more agents. This study proposes an alternative approach, called offline time-independent multi-agent path planning, without any timing assumptions. In any order of agent actions at execution, such plannings ensure that agents eventually reach their destinations. We show the advantages of the proposal through simulation with stochastic delays of agents' moves and demo with actual robots.

Anxiety disorder is a common mental disorder that affects people's daily life, and causes problems in social activities. Traditionally, doctors have relied on questioning symptoms, observing patients, and even using questionnaires to diagnose and treat their patients, lacking a knowledge framework for detailed and rapid diagnosis. Therefore, an easy-to-use expert system to diagnose anxiety would be very helpful to mental doctors, thus reducing doctors’ workload. SWI-Prolog is one of the popular expert system programming languages for building expert systems. Based on Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM–5) which is a common and useful guidebook for diagnosing mental disorders, this study designed and implemented an expert system for diagnosing anxiety disorders using SWI-Prolog and Java. This study also briefly introduces the basic syntax of SWI-Prolog, and explains how to connect Java and SWI-Prolog, in order to provide better visualization for users.

Since interpretations of statutes develop through time, PROLEG, as a computational representation of Japanese Civil Code with rules and exceptions, also requires revisions to update represented interpretations. This paper describes effects of revisions using a responsible case assignment proposed in previous works. Based on analysis of these effects, PROLEG can remind a user to check for non-trivial effects that the user might unintentionally make during the revision.

The microbiome drifting through the upper stream to downstream are considered to be affected by innumerable natural or artificial factors. Caching up the changing of microbiome from these factors has the great importance for social to make the novel index of environment pollution and health safety. Since DNA sequencing technology improved these days, the number of studying river metagenome are increasing and the feature of changing river microbiome are surveyed. However, the predictable models for river microbiome are still not understandable yet.In this study, we found the effective factors of microbiome from two river, Tama river and Sinos river, located from nature-close to city-close area and examined the possibility of constructing predictable models for river microbiome by comparing with each river.Our result showed that both Tama river and Sinos river preserved the microbiome in upper stream through the time, and in downstream of Tama river has the stable microbiome, and microbiome of Sinos river indicate that microbiome shift by longitudinal water flow.

Objective: Show bot behaviors on Twitter during the COVID-19 infodemic. In this paper, we examined the roles of bots in the case of the COVID-19 infodemic and the diffusion of non-credible information such as ``5G'' and ``Bill Gates'' conspiracy theories and ``Trump'' and ``WHO'' related contents by analyzing retweet networks and retweeted items. We collected more than 2 million COVID-19 related tweets and used Botometer to identify bots and human users in the retweet network filled with conspiracy theories emerged in the COVID-19 infodemic. We then labelled users as many as possible by examining published credible and non-credible URLs. In addition, we examined the bots and human behavior by counting the frequency of retweeting and finally revealed the top popular words list of human and bots by calculating TF-IDF. We show the segregated topology of their retweet networks, which indicates that right-wing self-media accounts and conspiracy theorists may lead to this opinion cleavage, while malicious bots might favor amplification of the diffusion of non-credible information. Although the basic influence of information diffusion could be larger in human users than bots, the effects of bots are non-negligible under an infodemic situation.<gdiv></gdiv>

One of the reasons why stock prices fluctuate greatly is because of IR announcements, which are information for investors, and news reports on events that are closely related to the companies. When a stock price change occurs, news sites for investors may report the stock price change and the reason for the change. However, such articles report only a certain portion of overall events that are closely related to reasons for stock price changes. Thus, in order to provide investors with information on those reasons for stock price changes, it is necessary to develop a system to collect information on events that could be closely related to the stock price changes of certain companies from the Internet. As the first step towards developing such a system, this paper takes an approach of employing a BERT-based machine reading comprehension model, which extracts reasons for stock price changes from news reports on stock price changes. Those extracted reasons are intended to be further used to train a system to collect information on events that could be closely related to the stock price changes of certain companies from the Internet.

Walkability analyses have recently gained attention for economic, health, and environmental reasons. We find clusters of areas with similar walkability profiles among the train stations of central Tokyo. First, we use a breadth-first search algorithm on the road network to determine the walkable areas within 5, 10, and 15 minutes of each station. We then collect the establishments within 50m of any traversed edge. We perform three analyses: (1) classifying regions by the numbers of stores of each type, (2) recursive feature selection and reclassification, and (3) scoring areas by their specialization in one of the categories. We find that classification without feature selection produces more useful results, and in both cases the <15 minute isochrones yield the best results. We also achieve realistic specialization results. These methods can be broadened to identify regions that are over- and under-serviced by amenities with impacts for both policy and business planning.

Neural Fuzzy Repair (NFR) system enables an NMT system to improve translation accuracy using similar translations searched from Translation Memory. This thesis compared edit distance and sentence-BERT (SBERT) as the similarity measures used in the search for similar translations, and showed that SBERT outperformed edit distance in the case of small corpus sizes. This thesis also studied a method to automatically select the most appropriate translation from more than one candidates. Compared to the naive method based on the number of tokens, the method based on the inner product of SBERT's sentence embedding achieved significant improvements. These results prove the effectiveness of the SBERT-based approach in the NFR system.

For dealing with time series forecasting problems, a machine learning method with a supervised learning algorithm can be considered as an efficient alternative tool. A long short-term memory (LSTM) approach, which is an advanced deep learning model, is considered. This study applied the LSTM method using a stochastic gradient descent with momentum, an adaptive moment estimation (Adam), a root mean square propagation algorithms for forecasting a chaotic time series problem (i.e. Mackey-Glass time series problem). This study also compared the results obtained using the LSTM method with those of obtained using a back-propagation neural network (BPNN) with a scaled conjugate gradient algorithm. Experimental results show that the LSTM approach with the Adam algorithm can be used efficiently to predict the pattern of the chaotic time series, and that the best results found by using the LSTM method and the BPNN are identical. Future work will use the LSTM approach for solving stock price prediction problems in the real-world.

Lately, researchers have used neural networks to solve partial differential equations (PDEs), enabling the mesh-free method for scientific computation. Unfortunately, the network performance drops when encountering unseen data points and a high nonlinearity domain. To improve the generalizability, we introduce the novel approach of employing the multi-task learning technique, called the uncertainty-weighting loss, in the context of learning PDE solutions. The multi-task scheme exploits the benefits of learning shared representations, controlled by cross-stitch modules, between multiple related PDEs. An auxiliary PDE is obtainable by varying the PDE parameterization coefficient, to generalize better on the original PDE. Letting the network pay closer attention to the high nonlinearity domain regions that are more challenging to learn, we also propose adversarial training for generating supplementary high-loss samples. In the experiment, our proposed method is found to be effective and reduce the error on the unseen data points as compared to the previous approaches.

For most board games, players commonly learn to increase strengths by following the opening moves played by experts, usually in the first stage of playing. In the past, many efforts have been made to use game-specific knowledge to construct opening books. Recently, DeepMind developed AlphaZero (2017) that can master game playing without domain knowledge. In this paper, we present an approach based on AlphaZero to constructing an opening book. To demonstrate the approach, we use a Connect6 program trained based on AlphaZero for evaluating positions, and then expand the opening game tree based on a job-level computing algorithm, called JL-UCT (job-level Upper Confidence Tree), developed by Wu et al. (2013) and Wei et al. (2015). In our experiment, the strengths of the Connect6 programs using this opening book are significantly improved, namely, the one with the opening book has a win rate of 65% against the one without using the book. In addition, the one without opening lost to Polygames in the Connect6 tournament of TCGA 2020 competitions, while the one with opening won against Polygames in TAAI and Computer Olympiad competitions later in 2020.