Timezone: »
This paper proposes a novel boosting algorithm called VadaBoost which is motivated by recent empirical Bernstein bounds. VadaBoost iteratively minimizes a cost function that balances the sample mean and the sample variance of the exponential loss. Each step of the proposed algorithm minimizes the cost efficiently by providing weighted data to a weak learner rather than requiring a brute force evaluation of all possible weak learners. Thus, the proposed algorithm solves a key limitation of previous empirical Bernstein boosting methods which required brute force enumeration of all possible weak learners. Experimental results confirm that the new algorithm achieves the performance improvements of EBBoost yet goes beyond decision stumps to handle any weak learner. Significant performance gains are obtained over AdaBoost for arbitrary weak learners including decision trees (CART).
Author Information
Pannagadatta K Shivaswamy (Cornell University)
Tony Jebara (Spotify)
More from the Same Authors
-
2019 Poster: A New Distribution on the Simplex with Auto-Encoding Applications »
Andrew Stirn · Tony Jebara · David Knowles -
2015 Workshop: Learning and privacy with incomplete data and weak supervision »
Giorgio Patrini · Tony Jebara · Richard Nock · Dimitrios Kotzias · Felix Xinnan Yu -
2014 Poster: Clamping Variables and Approximate Inference »
Adrian Weller · Tony Jebara -
2014 Poster: Making Pairwise Binary Graphical Models Attractive »
Nicholas Ruozzi · Tony Jebara -
2014 Spotlight: Making Pairwise Binary Graphical Models Attractive »
Nicholas Ruozzi · Tony Jebara -
2014 Oral: Clamping Variables and Approximate Inference »
Adrian Weller · Tony Jebara -
2013 Poster: A multi-agent control framework for co-adaptation in brain-computer interfaces »
Josh S Merel · Roy Fox · Tony Jebara · Liam Paninski -
2013 Poster: Adaptive Anonymity via $b$-Matching »
Krzysztof M Choromanski · Tony Jebara · Kui Tang -
2013 Spotlight: Adaptive Anonymity via $b$-Matching »
Krzysztof M Choromanski · Tony Jebara · Kui Tang -
2012 Workshop: Log-Linear Models »
Dimitri Kanevsky · Tony Jebara · Li Deng · Stephen Wright · Georg Heigold · Avishy Carmi -
2012 Poster: Majorization for CRFs and Latent Likelihoods »
Tony Jebara · Anna Choromanska -
2012 Spotlight: Majorization for CRFs and Latent Likelihoods »
Tony Jebara · Anna Choromanska -
2011 Poster: Learning a Distance Metric from a Network »
Blake Shaw · Bert Huang · Tony Jebara -
2008 Workshop: Analyzing Graphs: Theory and Applications »
Edo M Airoldi · David Blei · Jake M Hofman · Tony Jebara · Eric Xing -
2008 Poster: Relative Margin Machines »
Pannagadatta K Shivaswamy · Tony Jebara -
2008 Session: Oral session 8: Physics and High Order Statistics »
Tony Jebara -
2007 Poster: Density Estimation under Independent Similarly Distributed Sampling Assumptions »
Tony Jebara · Yingbo Song · Kapil Thadani -
2007 Spotlight: Density Estimation under Independent Similarly Distributed Sampling Assumptions »
Tony Jebara · Yingbo Song · Kapil Thadani -
2007 Spotlight: Learning Monotonic Transformations for Classification »
Andrew G Howard · Tony Jebara -
2007 Poster: Learning Monotonic Transformations for Classification »
Andrew G Howard · Tony Jebara -
2006 Poster: An EM Algorithm for Localizing Multiple Sound Sources in Reverberant Environments »
Michael Mandel · Daniel P Ellis · Tony Jebara -
2006 Poster: Gaussian and Wishart Hyperkernels »
Risi Kondor · Tony Jebara