Dissertation > Excellent graduate degree dissertation topics show
Online Learning and Its Implementation in Intelligent Traffic and Finantial Market
Author: ZhouZuo
Tutor: ZhangJunPing
School: Fudan University
Course: Computer Software and Theory
Keywords: Machine Learning Online Learning Perceptron Precise Parking Portfolio regret
CLC: U495;F830
Type: Master's thesis
Year: 2011
Downloads: 90
Quote: 1
Read: Download Dissertation
Abstract
|
Online learning is one of the hot research direction in recent years, the field of machine learning , increasingly widespread attention because of its highly consistent with a variety of problems in the real world . Have a natural advantage for new information to solve today 's social life constantly explosive growth and the continuous accumulation of old information that redundant processing . When the research on the topic of online learning is still at the theoretical level , the main concern is how solving and algorithms from online convex optimization perspective theoretical error bounds . The selected highly value a few questions online learning , including : 1 ) online algorithm nuclear methods and how to avoid the the growing nuclear matrix ; 2) regression algorithm is Online Boosting the performance advantages of the regression technique , and train accurate parking problems ; 3 ) how to take advantage of online algorithm design online portfolio strategy , how to use the characteristics of the fluctuations of the market itself to improve the online portfolio strategy performance . We propose the following improved algorithms and assess their performance : ( ?) Investigated several kernel-based Perceptron -like algorithm , and use of support vector machine soft margin improved a better performance Projectron algorithm , improved the SMProj algorithm to improve performance while also reducing the size of the support vector set . Good guide for the classification problems in real life . ( ? ) For the simplified model train parking investigated of several regression algorithm performance on the model , and GP Boosting algorithm proposed online method . Through the test confirmed the feasibility and advantages of the of Online Boosting regression algorithm in the train on the parking problem . (?) In the Anticor algorithm inspired , in-depth study proved that the mechanism of the algorithm in the portfolio , at the same time proposed algorithm doubt the improved CorrReg algorithm . Not only through the mutual relations of the market volatility and asset predict the direction of the flow of funds , and the use of internal regret online algorithms to quantify the number of the flow of capital . Making algorithm richer interpretability .
|
Related Dissertations
- Construction and improvement of basic education reform process evaluation platform,G521
- Implementation Study of evaluation method of the high school IT courses,G633.67
- CMC students online learning environment Evaluation,G434
- Logic-based Web service composition Petri net modeling and analysis,TP393.09
- Ontology -based Semantic Web service matching and composition method,TP393.09
- The Research about the Collection Asset Management Plan’s Operation in China Security Market,F832.51
- Semi-supervised Learning and Active Learning of Sentiment Classification Coupled with Domain Knowledge,TP181
- Application Research in Stock Selection and Optimization Strategies of Portfolio Investment,F830.91
- Capsule endoscopy and endoscopic image portable receiver system bleeding Recognition Algorithm,TP391.41
- Application of Data Mining in Email Anti-Spam System,TP393.098
- Graph model based on statistical and machine learning algorithm and its application of certain,TP181
- Risk Control Based on Copula loan portfolio optimization model,F224
- SVM Based on SIFT and scene classification,TP391.41
- Trait volatility on stock returns impact analysis,F224
- Default correlation based on the degree of concentration risk control method,F830.5
- Random Forests Feature Selection,TP311.13
- Research of Protein-Protein Interaction Extraction Based on Rich Feature and Multiple Kernels Learning,Q51
- Brain Connectivity Pattern Analysis Based on DTI,R445.2
- A Comparative Study of Four English Versions of Bai Juyi’s Chang Hen Ge,I046
- The Research on Consumer’s Right to Regret,D923.8
CLC: > Economic > Fiscal, monetary > Finance, banking > Finance, banking theory
© 2012 www.DissertationTopic.Net Mobile
|