MINING EXCEPTIONAL ACTIVITY PATTERNS IN MICROSTRUCTURE DATA
Contact CMCRC for this article Author(s): Yuming Ou, Longbing Cao, Chao Luo and Li Liu
DECISION TREES FOR IMBALANCED DATA SETS
Contact CMCRC for this article Author(s): Wei Liu, Sanjay Chawla
OUTLIER MINING ON MULTIPLE TIME SERIES DATA IN STOCK MARKET
Contact CMCRC for this article Author(s): Chao Luo, Yanchang Zhao, Dan Luo, Yuming Ou, Li Liu
TIKHONOV OR LASSO REGULARIZATION: WHICH IS BETTER AND WHEN. IN TOOLS WITH ARTIFICIAL INTELLIGENCE
pages 795–802. IEEE, 2013. It is well known that supervised learning problems with ℓ1 (Lasso) and ℓ2 (Tikhonov or Ridge) regularizers will result in very different solutions. For example, the ℓ1 solution vector will be sparser and can potentially be used both for prediction and feature selection. However, given a data set it is often hard to determine
OWNERSHIP CONCENTRATION, INDUSTRIAL COMPETITION AND INFORMATION ASYMMETRY
Managers and major shareholders typically have an information advantage over outside investors when trading a company’s shares. Numerous studies have been done into the relation between ownership concentration, industry competition and information asymmetry, but rarely using Australian data. The output of this study is particularly relevant for regulators, as it examines conditions that might limit
LOCATION INDICATIVE WORDS AND HOW TO FIND THEM
Speaker: Bo Han (RedMarker/Kaplan Professional) Seminar Date: Tuesday November 28 12:00pm Brief abstract: Social media has become a gold mine of insights into the people, opinions and events. However, this massive amount of data also challenges the efficiency and effectiveness of existing machine learning algorithms. One way to approach this problem is “divide-and-conquer”. You can partition the data into
AN INTRODUCTION TO ACTIVE LEARNING, AND ITS APPLICATION TO BUSINESS COACHING
Speaker: Tom Osborn Seminar Date: Tuesday November 21 12:00pm Brief abstract: Active learning is a branch of semi-supervised machine learning for domains where data is expensive and where precise understanding is critical. Traditionally, active learning builds a model of a domain with a trade-off between exploration and exploitation probing of a domain to generate labelled data usefully. Earlier applications
GAMING IN FUTURES MARKET SUGGESTS MODEL COULD BE IMPROVED?
Evidence from UK interest rate futures market reveals that traders have been “drowning” the market with oversized orders, increasing their allocation under a pure pro-rata matching algorithm. The 2007 introduction of a time element to the order matching mechanism has modified the behaviour of traders. No longer is the order book drowned with orders which
FACING FEDERATION: STREAMLINING APPLICATIONS AND APPROVALS FOR CROSS-JURISDICTIONAL DATA LINKAGE IN AUSTRALIA.
Speaker: Dr Felicity Flack, PHRN Seminar Date: Tuesday October 17 12:00pm Brief abstract: The Population Health Research Network (PHRN) has been funded via the National Collaborative Research Infrastructure Strategy (NCRIS), in collaboration with universities, as well as state and territory governments, to develop national data linkage infrastructure for Australian researchers. In 2013-14 the PHRN Program Office consulted with the
HOW TECHNOLOGY CAN HELP OLDER PEOPLE AGING WELL
Speaker: Michael Sheng Seminar Date: Tuesday October 10 12:00pm Short Bio: Michael Sheng is a professor at Macquarie University, Sydney, since 1 January 2017. Dr. Michael Sheng was a full Professor and Deputy Head of the School of Computer Science at the University of Adelaide. Michael holds a PhD degree in computer science from the University of New South