CSML talks for this week

[Alex Shestopaloff]

Hi everyone,

We have three CSML talks this week!

Zacharie Naulet (Paris-Sud)
Thursday February 27th at 3:30 pm, LG.04

Title: 

Risk of the Least Squares Minimum Norm Estimator under the Spike Covariance Model

Abstract:

We study risk of the minimum norm linear least squares estimator in when the number of parameters $d$ depends on $n$, and $\frac{d}{n} \rightarrow \infty$. We assume that data has an underlying low rank structure by restricting ourselves to spike covariance matrices, where a fixed finite number of eigenvalues grow with $n$ and are much larger than the rest of the eigenvalues, which are (asymptotically) in the same order. We show that in this setting risk of minimum norm least squares estimator vanishes in compare to risk of the null estimator. We give asymptotic and non asymptotic upper bounds for this risk, and also leverage the assumption of spike model to give an analysis of the bias that leads to tighter bounds in compare to previous works.

Arthur Jacot (EPFL)
Friday February 28th at 11:00 am, Small Lecture Theatre

Title: 

Implicit Regularization of Random Feature Models

Abstract:

Random Feature (RF) models are used as efficient parametric approximations of kernel methods. We investigate, by means of random matrix theory, the connection between Gaussian RF models and Kernel Ridge Regression (KRR). For a Gaussian RF model with an arbitrary original ridge, the average (i.e. expected) RF predictor is close to a KRR predictor with a effective ridge, larger than the original, thus revealing the implicit regularization effect of finite RF sampling. Our proofs show the close relation between the effective ridge and the Stieltjes transform of generalized Wishart random matrices. Our setting also sheds a new light on the ridgeless case, where the celebrated double-descent phenomenon takes place.

Danielle Belgrave (Microsoft Research)
Friday February 28th at 3:30 pm, Small Lecture Theatre

To be announced.

 Best,

 -- Alex

[Alex Shestopaloff]

Reminder of Zacharie's talk later today.

[Yee Whye Teh]

Hi Alex, I'd like to meet with Arthur Jacot please. Maybe at 130-200pm.

-yw


::: Yee Whye Teh ::: Professor of Statistical Machine Learning :::
::: Oxford Statistics and DeepMind :::
http://mlcs.stats.ox.ac.uk/people/teh/ :::

> --
> Oxford Computational Statistics and Machine Learning
> Website: http://csml.stats.ox.ac.uk/
> GitHub: https://github.com/oxmlcs/
> Events: https://github.com/oxmlcs/ML_bazaar/wiki/Machine-Learning-Bazaar-Schedule
> ---
> You received this message because you are subscribed to the Google Groups "OxCSML Public Announcements" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to oxcsml-publi...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/oxcsml-public/CACin1tL2QsyTBC6WX790bmVNFH_kufHpw18epvPSt0wmq9Thsg%40mail.gmail.com.

[Alex Shestopaloff]

Reminder of Arthur's and Danielle's talks later today.

The updated abstract for Danielle's talk is:

Machine Learning for personalised healthcare

Machine learning advances are opening new routes to more precise healthcare, from the discovery of disease subtypes for stratified interventions to the development of personalised interactions supporting self-care between clinic visits. This offers an exciting opportunity for machine learning techniques to impact healthcare in a meaningful way. In this talk, I will present recent work on probabilistic graphical modelling frameworks to enable a more personalised approach to healthcare, whereby information can be aggregated from multiple sources within a unified modelling framework. The work presented will be motivated within the clinical contexts of asthma, allergic diseases and mental health.

On Mon, Feb 24, 2020 at 6:52 AM Alex Shestopaloff <shestopa...@gmail.com> wrote: