The Upward Bias in Measures of Information Derived from Limited Data Samples
Abstract
Extracting information measures from limited experimental samples, such as those normally available when using data recorded in vivo from mammalian cortical neurons, is known to be plagued by a systematic error, which tends to bias the estimate upward. We calculate here the average of the bias, under certain conditions, as an asymptotic expansion in the inverse of the size of the data sample. The result agrees with numerical simulations, and is applicable, as an additive correction term, to measurements obtained under such conditions. Moreover, we discuss the implications for measurements obtained through other usual procedures.
Cite
Text
Treves and Panzeri. "The Upward Bias in Measures of Information Derived from Limited Data Samples." Neural Computation, 1995. doi:10.1162/NECO.1995.7.2.399Markdown
[Treves and Panzeri. "The Upward Bias in Measures of Information Derived from Limited Data Samples." Neural Computation, 1995.](https://mlanthology.org/neco/1995/treves1995neco-upward/) doi:10.1162/NECO.1995.7.2.399BibTeX
@article{treves1995neco-upward,
title = {{The Upward Bias in Measures of Information Derived from Limited Data Samples}},
author = {Treves, Alessandro and Panzeri, Stefano},
journal = {Neural Computation},
year = {1995},
pages = {399-407},
doi = {10.1162/NECO.1995.7.2.399},
volume = {7},
url = {https://mlanthology.org/neco/1995/treves1995neco-upward/}
}