Gradient Boosted Regression Trees (GBRT) are the current state-of-the-art learning paradigm for machine learned websearch ranking — a domain notorious for very large data sets. ...
Stephen Tyree, Kilian Q. Weinberger, Kunal Agrawal...
This paper describes revised content-based search experiments in the context of TRECVID 2003 benchmark. Experiments focus on measuring content-based video retrieval performance wi...
The particle swarm optimization algorithm was showed to converge rapidly during the initial stages of a global search, but around global optimum, the search process will become ve...
Jing-Ru Zhang, Jun Zhang, Tat-Ming Lok, Michael R....
The number of required hidden units is statistically estimated for feedforward neural networks that are constructed by adding hidden units one by one. The output error decreases w...
— A heuristic is proposed to address free parameter selection for Support Vector Machines, with the goals of improving generalization performance and providing greater insensitiv...