Creating optimal logic mining is strongly dependent on how the learning data are structured. Without optimal data structure, intelligence systems integrated into logic mining, such as an artificial neural network, tend to converge to suboptimal solution. This paper proposed a novel logic mining that integrates supervised learning via association analysis to identify the most optimal arrangement with respect to the given logical rule. By utilizing Hopfield neural network as an associative memory to store information of the logical rule, the optimal logical rule from the correlation analysis will be learned and the corresponding optimal induced logical rule can be obtained. In other words, the optimal logical rule increases the chances for the logic mining to locate the optimal induced logic that generalize the datasets. The proposed work is extensively tested on a variety of benchmark datasets with various performance metrics. Based on the experimental results, the proposed supervised logic mining demonstrated superiority and the least competitiveness compared to the existing method.

/pdf/supervised-learning-perspective-in-logic-mining-2cd5g34n.pdf

Supervised Learning Perspective in Logic Mining

Determination of ductile fracture properties of 16MND5 steels under varying constraint levels using machine learning methods

This systematic literature review analyses the recent advances of machine learning and deep learning in finance. The study considers six financial domains: stock markets, portfolio management, cryptocurrency, forex markets, financial crisis, bankruptcy and insolvency. We provide an overview of previously proposed techniques in these areas by examining 126 selected articles across 44 reputed journals. The main contributions of this review include an extensive examination of data characteristics and features used for model training, evaluation of validation approaches, and model performance addressing each financial problem. A systematic literature review methodology, PRISMA, is used to carry out this comprehensive review. The study also analyses bibliometric information to understand the current status of research focused on machine learning in finance. The study finally points out possible research directions which might lead to new inquiries in machine learning and finance. 

Financial applications of machine learning: A literature review

Rapid detection of thiabendazole in food using SERS coupled with flower-like AgNPs and PSL-based variable selection algorithms

This study aims to develop a foreign exchange trading model with a new leading indicator based on cross-currency swap (CCS) and interest rate swap (IRS) rates. We derive the correlation dimension from the chaos analysis, use Neural Networks (NNs) to optimize foreign exchange trading, and then compare the performance of the NNs model with XGBoost and Logistic Regression model (LR). The CCS-IRS basis is the spread between the cross-currency swap (CCS) rate and interest rate swap (IRS) rate and is often used as an indicator for foreign exchange soundness along with the CDS premium in Korea. Based on the phenomenon in which the USDKRW exchange rate rises (falls) when the CCS-IRS basis widens (tightens), this study proposes to use the CCS-IRS basis to build a foreign exchange trading model of the USDKRW exchange rate. The experimental results show that our NNs trading model of the USDKRW exchange rate outperforms XGBoost and LR model, generating an annual return of 4.888%∼8.464%, whereas the buy-and-hold and sell-and-hold strategies in the USDKRW exchange rate market produces an annual return of 2.2% and −1.15%, respectively. The foreign exchange trading model established in this paper not only enables foreign exchange traders to obtain sustainable economic profits, but also contributes to the efficiency of the financial market, thereby contributing to sustainable economic growth. 

Neural network foreign exchange trading system using CCS-IRS basis: Empirical evidence from Korea

Data selection to avoid overfitting for foreign exchange intraday trading with machine learning

The valuation of discounted cash flow (DCF) is one of the most appropriate methods to measure a company’s value, and it is also a fundamental method according to Warren Buffett to value a stock for investment. DCF analysis requires the estimations of the future free cash flow (FCF) and the weighted average cost of capital. However, these factors are often subject to an appraiser’s bias. To obtain a more objective DCF analysis, in this study, we develop a machine learning mechanism: imitation learning, which encompasses a teacher to demonstrate state-action pairs. To facilitate imitation learning, we adopt a guided policy search with the chronological order of a Levy distribution. This distribution can cover the stages of growth, decay, and long fat tail that correspond to a business life cycle. In this way, learning the trajectory of FCF can optimize the fitness between the FCF and the chronological order of a Levy distribution. The fitness achieved allows us to monitor the data by fitting the chronological order for the correct stage. The results show that regarding the Sharpe ratio and precision of DCF valuation, our approach outperforms the popular classification models -random forest and long short-term memory network. Code is available at https://github.com/YuanLongPeng/DCFValuation. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —Appraisers own knowledge for valuation of stocks to project FCFs and WACC. The knowledge has a broad spectrum, such as prospect of economic growth and competition in the industry. Threat of substitute products. This study proposed a method to automate the process of valuation by learning FCFs and WACC from data. The result of valuation is reproducible and free of the bias of appraisers. 

Valuation of Stocks by Integrating Discounted Cash Flow With Imitation Learning and Guided Policy

Following the rise of internet, multimedia applications have become popular and been expanded rapidly. How to retrieve the user-specified video content is an important issue. Since the query varies from user to user for the same content, one turn search usually cannot find the satisfying answer. Therefore, we extend the typical single-turn multimodal video clip retrieval to a multi-turn system. To pull this off, video data are transformed into feature vectors. At each turn, the user inputs a query sentence to the system, and the system will output six candidate video clips accordingly. The system recommends video clips based on the results of a clustering algorithm and video retrievals benefit from the similarity feedback of previous turn. The results show that the proposed method outperformed the single-turn retrieval on the metrics of recall and the similarity.

Yuan-Long Peng

Papers

Data selection to avoid overfitting for foreign exchange intraday trading with machine learning

Valuation of Stocks by Integrating Discounted Cash Flow With Imitation Learning and Guided Policy

Multi-turn Query with Similarity Feedback Facilitates Multimodal Video Clip Retrieval