3
$\begingroup$

I am doing a PhD in functional analysis, particularly Banach space theory. Thinking of my future prospect, I would like to venture myself into statistics.

However, as I do not want to 'waste' my background, I would like to search for statistics or machine learning which requires functional analysis to understand its underlying concepts.

I found Bickel et al. Efficient and Adaptive Estimation for Semiparametric Models, Tsiatis's Semiparametric Theory and Missing Data and Small et al. Hilbert Space Methods in Probability and Statistical Inference.

Note that I know functional analysis is required for stochastic calculus, theoretical probability, browninan motion, etc. But I am more interested in statistics.

I also found that similar question was posted reddit.

It would be good if someone can suggest some statistics books that require functional analysis background.

$\endgroup$
  • $\begingroup$ @kjetilbhalvorsen Two of your links above are assuming that I want to study functional analysis because I need it in statistics. But my situation is the converse, that is, I have background in functional analysis, but I want to venture into statistics. $\endgroup$ – Idonknow Jul 25 at 13:49
  • $\begingroup$ Have a look at Gaussian process es for machine learning book and support vector machines (reproducing kernel Hilbert spaces) $\endgroup$ – seanv507 Aug 24 at 16:44
3
$\begingroup$

I would suggest looking through the functional analysis literature for papers relating to statistics and machine learning. Books aren't where the cutting-edge results are found, and your background will allow you to understand the functional analysis papers.

In a few minutes on the website for what appears to be a strong functional analysis journal, I found the following papers.

Newton, Nigel J. "An infinite-dimensional statistical manifold modelled on Hilbert space." Journal of Functional Analysis 263.6 (2012): 1661-1681. https://www.sciencedirect.com/science/article/pii/S0022123612002376

Jenčová, Anna. "A construction of a nonparametric quantum information manifold." Journal of Functional Analysis 239.1 (2006): 1-20. https://www.sciencedirect.com/science/article/pii/S0022123606000644

Hangelbroek, Thomas, and Amos Ron. "Nonlinear approximation using Gaussian kernels." Journal of Functional Analysis 259.1 (2010): 203-219. https://www.sciencedirect.com/science/article/pii/S0022123610000467

$\endgroup$
  • $\begingroup$ All those papers are from a reputable journal: journal of functional analysis. Nonetheless, thanks for suggesting. I didn't realize that there are papers of functional analysis on statistics. $\endgroup$ – Idonknow Jul 25 at 13:31
2
$\begingroup$

A graduate level book on Statistics and Functional Analysis. This post Is functional analysis and hilbert spaces useful in machine learning? If so, how? gives some hints&references.

$\endgroup$
  • $\begingroup$ How stat-y is that book? It seems to be oriented towards probability and stochastic processes (hence the title) rather than statistical inference. $\endgroup$ – Dave Jul 25 at 14:43
2
$\begingroup$

Simovici 2018, Mathematical Analysis for Machine Learning and Data Mining

This book might be subtitled "an introduction to machine learning for functional analysts".

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.