Nicolaou A, Dey S, Christlein V, Maier A, Karatzas D (2019)
Publication Language: English
Publication Type: Conference contribution
Publication year: 2019
Publisher: Springer Verlag
Book Volume: 11455 LNCS
Pages Range: 71-82
Conference Proceedings Title: RRPR 2018: Reproducible Research in Pattern Recognition
ISBN: 9783030239862
URI: https://arxiv.org/pdf/1806.07171.pdf
DOI: 10.1007/978-3-030-23987-9_5
Embedding data into vector spaces is a very popular strategy of pattern recognition methods. When distances between embeddings are quantized, performance metrics become ambiguous. In this paper, we present an analysis of the ambiguity quantized distances introduce and provide bounds on the effect. We demonstrate that it can have a measurable effect in empirical data in state-of-the-art systems. We also approach the phenomenon from a computer security perspective and demonstrate how someone being evaluated by a third party can exploit this ambiguity and greatly outperform a random predictor without even access to the input data. We also suggest a simple solution making the performance metrics, which rely on ranking, totally deterministic and impervious to such exploits.
APA:
Nicolaou, A., Dey, S., Christlein, V., Maier, A., & Karatzas, D. (2019). Non-deterministic Behavior of Ranking-Based Metrics When Evaluating Embeddings. In Hugues Talbot, Pascal Monasse, Bertrand Kerautret, Miguel Colom, Daniel Lopresti (Eds.), RRPR 2018: Reproducible Research in Pattern Recognition (pp. 71-82). Beijing, CN: Springer Verlag.
MLA:
Nicolaou, Anguelos, et al. "Non-deterministic Behavior of Ranking-Based Metrics When Evaluating Embeddings." Proceedings of the 2nd International Workshop on Reproducible Research in Pattern Recognition, RRPR 2018, Beijing Ed. Hugues Talbot, Pascal Monasse, Bertrand Kerautret, Miguel Colom, Daniel Lopresti, Springer Verlag, 2019. 71-82.
BibTeX: Download