Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Time-series clustering - A decade review

Contents: J. Gale, Preface. Part I:Radical Constructivism and Social Constructionism. E. von Glasersfeld, A Constructivist Approach to Teaching. K.J. Gergen, Social Construction and the Educational Process. J. Shotter, In Dialogue: Social Constructionism and Radical Constructivism. J. Richards, Construct[ion/iv]ism: Pick One of the Above. Part II:Information-Processing Constructivism and Cybernetic Systems. F. Steier, From Universing to Conversing: An Ecological Constructionist Approach to Learning and Multiple Description. R.J. Spiro, P.J. Feltovich, M.J. Jacobson, R.L. Coulson, Cognitive Flexibility, Constructivism, and Hypertext: Random Access Instruction for Advanced Knowledge Acquisition in Ill-Structured Domains. K. Tomm, Response to Chapters by Spiro et al. and Steier. P.W. Thompson, Constructivism, Cybernetics, and Information Processing: Implications for Technologies of Research on Learning. Part III:Social Constructivism and Sociocultural Approaches. H. Bauersfeld, The Structuring of the Structures: Development and Function of Mathematizing as a Social Practice. J.V. Wertsch, C. Toma, Discourse and Learning in the Classroom: A Sociocultural Approach. C. Konold, Social and Cultural Dimensions of Knowledge and Classroom Teaching. J. Confrey, How Compatible Are Radical Constructivism, Sociocultural Approaches, and Social Constructivism? Analysis and Synthesis I: Alternative Epistemologies. M.H. Bickhard, World Mirroring Versus World Making: There's Gotta Be a Better Way. Part IV:Alternative Epistemologies in Language, Mathematics, and Science Education. R. Duit, The Constructivist View: A Fashionable and Fruitful Paradigm for Science Education Research and Practice. G.B. Saxe, From the Field to the Classroom: Studies in Mathematical Understanding. N.N. Spivey, Written Discourse: A Constructivist Perspective. T. Wood, From Alternative Epistemologies to Practice in Education: Rethinking What It Means to Teach and Learn. E. Ackermann, Construction and Transference of Meaning Through Form. D. Rubin, Constructivism, Sexual Harassment, and Presupposition: A (Very) Loose Response to Duit, Saxe, and Spivey. Part V:Alternative Epistemologies in Clinical, Mathematics, and Science Education. E. von Glasersfeld, Sensory Experience, Abstraction, and Teaching. R. Driver, Constructivist Approaches to Science Teaching. T. Wood, P. Cobb, E. Yackel, Reflections on Learning and Teaching Mathematics in Elementary School. P. Lewin, The Social Already Inhabits the Epistemic: A Discussion of Driver Wood, Cobb, and Yackel and von Glasersfeld. J. Becker, M. Varelas, Assisting Construction: The Role of the Teacher in Assisting the Learner's Construction of Preexisting Cultural Knowledge. E.H. Auerswald, Shifting Paradigms: A Self-Reflective Critique. Analysis and Synthesis II: Epsitemologies in Education. P. Ernest, The One and the Many. Analysis and Synthesis III: Retrospective Comments and Future Prospects. L.P. Steffe, Alternative Epistemologies: An Educator's Perspective. J. Gale, Epilogue.

教育中的建构主义 = Constructivism in Education

Financial time series forecasting with deep learning : A systematic literature review: 2005–2019

Forecasting electricity consumption: A comparison of regression analysis, neural networks and least squares support vector machines

In this paper, the development of knowledge management (KM) was surveyed, using a literature review and classification of articles from 1995 to 2004. With a keyword index and article abstract, we explored how KM performance evaluation has developed during this period. Based on a scope of 108 articles from 80 academic KM journals (retrieved from six online databases), we surveyed and classified methods of KM measurement, using the following eight categories: qualitative analysis, quantitative analysis, financial indicator analysis, non-financial indicator analysis, internal performance analysis, external performance analysis, project-orientated analysis and organizationorientated analysis, together with their measurement matrices for different research and problem domains. Future development directions for KM performance evaluation are presented in our discussion. They include: (1) KM performance measurements have tended towards expertise orientation, while evaluation development is a problemorientated domain; (2) different information technology methodologies, such as expert systems, knowledge-based systems and case-based reasoning may be able to evaluate KM as simply another methodology; (3) the ability to continually change and obtain new understanding is the driving power behind KM methodologies, and should be the basis of KM performance evaluations in the future.

https://ir.nctu.edu.tw:443/bitstream/11536/12822/1/000236053800003.pdf

Knowledge management performance evaluation: a decade review from 1995 to 2004

A novel financial time-series analysis method based on deep learning technique is proposed in this paper. In recent years, the explosive growth of deep learning researches have led to several successful applications in various artificial intelligence and multimedia fields, such as visual recognition, robot vision, and natural language processing. In this paper, we focus on the time-series data processing and prediction in financial markets. Traditional feature extraction approaches in intelligent trading decision support system are used to applying several technical indicators and expert rules to extract numerical features. The major contribution of this paper is to improve the algorithmic trading framework with the proposed planar feature representation methods and deep convolutional neural networks (CNN). The proposed system is implemented and benchmarked in the historical datasets of Taiwan Stock Index Futures. The experimental results show that the deep learning technique is effective in our trading simulation application, and may have greater potentialities to model the noisy financial data and complex social science problems. In the future, we expected that the proposed methods and deep learning framework could be applied to more innovative applications in the next financial technology (FinTech) generation.

Financial Time-Series Data Analysis Using Deep Convolutional Neural Networks

The knowledge-based economy is coming, and knowledge management (KM) has rapidly disseminated in academic circles as well as in the business world. While an increasing number of companies have launched into knowledge management initiatives, a large proportion of these initiatives are limited to a technical focus. The problem with this type of focus is that it excludes and neglects the true potential benefits that can be derived from knowledge management. This paper develops a new metric, knowledge management performance index (KMPI), for evaluating the performance of a firm in its KM at a point in time. We therefore suggest that a KMPI can be used to determine KM activities from the following perspectives: knowledge creation, knowledge conversion, knowledge circulation and knowledge completion. When KM activities efficiency is increased, KMPI will also be expanded, enabling firms to become knowledge intensive. This paper makes three important contributions: (1) it provides a formal theoretical grounding for the validity of the Black-Scholes model that might be applied to KM; (2) it proposes a measurement framework to enable knowledge assets to be leveraged effectively and efficiently; and (3) it presents the first application of the Black-Scholes model that uses a real-world business situation involving KM as its test bed. The results prove the option pricing model can act as a measurement guideline to the whole range of KM activities.

https://ir.nctu.edu.tw:443/bitstream/11536/25143/1/000232107900005.pdf

Integrating option model and knowledge management performance measures: an empirical study

Machine learning methods such as fuzzy logic, neural networks and decision tree induction have been applied to learn rules, however they can get trapped into a local optimal. Based on the principle of natural evolution and global searching, a genetic algorithm is promising for obtaining better results. This article adopts the learning classifier systems (LCS) technique to provide a three-phase knowledge extraction methodology, which makes continues and instant learning while integrates multiple rule sets into a centralized knowledge base. This paper makes three important contributions: (1) it represents various rule sets that are derived from different sources and encoded as a fixed-length bit string in the knowledge encoding phase; (2) it uses three criteria (accuracy, coverage, and fitness) to select an optimal set of rules from a large population in the knowledge extraction phase; (3) it applies genetic operations to generate optimal rule sets in the knowledge integration phase. The experiments prove that the rule sets derived by the proposed approach is more accurate than other machine learning algorithm.

https://ir.nctu.edu.tw:443/bitstream/11536/12106/1/000236903700020.pdf

Integrating extended classifier system and knowledge extraction model for financial investment prediction: An empirical study

This research applies a market profile to establish an indicator to classify the correlation between the variation in price and value with the stock trends. The indicator and technical index are neural network architecture parameters that assist to extrapolate the market logic and knowledge rules that influence the TAIEX futures market structure via an integral assessment of physical quantities. To implement the theory of market profile on neural network architecture, this study proposes qualitative and quantitative methods to compute a market profile indicator. In addition, the indicator considers the variation and relevance between long-term and short-term trends by incorporating the long-term and short-term change in market in its calculation. An assessment of forecasting performance on different calculation approaches of market profile indicator and technical analysis is conducted to differentiate their accuracies and profitability. The experimental results show the qualitative market profile indicator outperforms the quantitative approach in a short-term forecast period. In contrast, the quantitative market profile indicator has a better trend-predicting ability, thus it is more effective in the long-term forecast period. The integration of market profile and technical analysis surpasses technical analysis as a neural network architecture parameter by effectively improving forecasting performance and profitability.

https://ir.nctu.edu.tw:443/bitstream/11536/24386/1/000335629500010.pdf

An-Pin Chen

Papers

Knowledge management performance evaluation: a decade review from 1995 to 2004

Financial Time-Series Data Analysis Using Deep Convolutional Neural Networks

Integrating option model and knowledge management performance measures: an empirical study

Integrating extended classifier system and knowledge extraction model for financial investment prediction: An empirical study

Applying market profile theory to forecast Taiwan Index Futures market