arXiv subjects

Sun-Chong Wang

Publications and source records attributed to Sun-Chong Wang.

4 recordsLinked to original sources

arXiv2017

Neural Machine Translation between Herbal Prescriptions and Diseases

The current study applies deep learning to herbalism. Toward the goal, we acquired the de-identified health insurance reimbursements that were claimed in a 10-year period from 2004 to 2013 in the National Health Insurance Database of Taiwan, the total number of reimbursement records equaling 340 millions. Two artificial intelligence techniques were applied to the dataset: residual convolutional neural network multitask classifier and attention-based recurrent neural network. The former works to translate from herbal prescriptions to diseases; and the latter from diseases to herbal prescriptions. Analysis of the classification results indicates that herbal prescriptions are specific to: anatomy, pathophysiology, sex and age of the patient, and season and year of the prescription. Further analysis identifies temperature and gross domestic product as the meteorological and socioeconomic factors that are associated with herbal prescriptions. Analysis of the neural machine transitional result indicates that the recurrent neural network learnt not only syntax but also semantics of diseases and herbal prescriptions.

cs.CL↗

arXiv2010

Statistical properties of agent-based models in markets with continuous double auction mechanism

Real world markets display power-law features in variables such as price fluctuations in stocks. To further understand market behavior, we have conducted a series of market experiments on our web-based prediction market platform which allows us to reconstruct transaction networks among traders. From these networks, we are able to record the degree of a trader, the size of a community of traders, the transaction time interval among traders and other variables that are of interest. The distributions of all these variables show power-law behavior. On the other hand, agent-based models have been proposed to study the properties of real financial markets. We here study the statistical properties of these agent-based models and compare them with the results from our web-based market experiments. In this work, three agent-based models are studied, namely, zero-intelligence (ZI), zero-intelligence-plus (ZIP) and Gjerstad-Dickhaut (GD). Computer simulations of variables based on these three agent-based models were carried out. We found that although being the most naive agent-based model, ZI indeed best describes the properties observed in real markets. Our study suggests that the basic ingredient to produce the observed properties from real world markets could in fact be the result of a continuously evolving dynamical system with basic features similar to the ZI model.

q-fin.TR↗

arXiv2010

Experimental evidence for the interplay between individual wealth and transaction network

We conduct a market experiment with human agents in order to explore the structure of transaction networks and to study the dynamics of wealth accumulation. The experiment is carried out on our platform for 97 days with 2,095 effective participants and 16,936 times of transactions. From these data, the hybrid distribution (log-normal bulk and power-law tail) in the wealth is observed and we demonstrate that the transaction networks in our market are always scale-free and disassortative even for those with the size of the order of few hundred. We further discover that the individual wealth is correlated with its degree by a power-law function which allows us to relate the exponent of the transaction network degree distribution to the Pareto index in wealth distribution.

physics.soc-ph↗

arXiv2005

Statistical properties of an experimental political futures market

A 24-hour exchange market was created on the Web to trade political futures contracts using fictitious money. In this online market, a political futures contract is a futures contract which matures on the election day with a liquidation price determined by the percentage of votes a candidate receives on the election day. Continuous double auctions were implemented as the system for order storage and price discovery. We drew market participants in the form of tournaments in which top traders won cash awards. Such a market was run, with about 400 registered traders, during the U.S. presidential election in November 2004 and Taiwan parliamentary election in December 2004. The experiments recorded transaction price, highest bid, lowest ask, and trading volume of each contract as a function of time. Despite the relatively small scale of the exchange, in terms of the number of participants and duration of the tournament, we report evidence for asymptotic power-law behaviors of the distributions of price returns, trading volumes, inter-transaction time intervals, and accumulated wealth that were found universal in real financial markets.

physics.soc-ph↗