Quantification in-the-wild: data-sets and baselines

• DOI: 10.48550/arxiv.1510.04811
• Publication Date: 2015-01-01
• Authors: Beijbom, Oscar; Hoffman, Judy; Yao, Evan; Darrell, Trevor; Rodriguez-Ramirez...; Gonzalez-Rivero, ...; Guldberg, Ove Hoe...
• Keywords: FOS: Computer and information sciences; Computer Science - Machine Learning; 0202 electrical engineering, electronic engineering, information engineering; 14. Life underwater; 02 engineering and technology; Machine Learning (cs.LG)

This report was prsented at the NIPS 2015 workshop on Transfer and Multi-Task Learning: Trends and New Perspectives. It is 4 pages + 1 page of references followed by a 6 page appendix<br/>Quantification is the task of estimating the class-distribution of a data-set. While typically considered as a parameter estimation problem with strict assumptions on the data-set shift, we consider quantification in-the-wild, on two large scale data-sets from marine ecology: a survey of Caribbean coral reefs, and a plankton time series from Martha's Vineyard Coastal Observatory. We investigate several quantification methods from the literature and indicate opportunities for future work. In particular, we show that a deep neural network can be fine-tuned on a very limited amount of data (25 - 100 samples) to outperform alternative methods.<br/>