Projects per year
Abstract
In highdimensional data, structured noise caused by observed and unobserved factors affecting multiple target variables simultaneously, imposes a serious challenge for modeling, by masking the often weak signal. Therefore, (1) explaining away the structured noise in multipleoutput regression is of paramount importance. Additionally, (2) assumptions about the correlation structure of the regression weights are needed. We note that both can be formulated in a natural way in a latent variable model, in which both the interesting signal and the noise are mediated through the same latent factors. Under this assumption, the signal model then borrows strength from the noise model by encouraging similar effects on correlated targets. We introduce a hyperparameter for the latent signaltonoise ratio which turns out to be important for modelling weak signals, and an ordered infinitedimensional shrinkage prior that resolves the rotational unidentifiability in reducedrank regression models. Simulations and prediction experiments with metabolite, gene expression, FMRI measurement, and macroeconomic time series data show that our model equals or exceeds the stateoftheart performance and, in particular, outperforms the standard approach of assuming independent noise and signal models.
Original language  English 

Pages (fromto)  135 
Number of pages  35 
Journal  Journal of Machine Learning Research 
Volume  17 
Publication status  Published  1 Jun 2016 
MoE publication type  A1 Journal articlerefereed 
Keywords
 Bayesian reducedrank regression
 Latent signaltonoise ratio
 Latent variable models
 Multipleoutput regression
 Nonparametric Bayes
 Shrinkage priors
 Structured noise
 Weak effects
Fingerprint
Dive into the research topics of 'Multiple output regression with latent noise'. Together they form a unique fingerprint.Projects
 3 Finished

Interactive machine learning from multiple biodata sources
Kaski, S. & Filstroff, L.
01/01/2016 → 31/08/2021
Project: Academy of Finland: Other research funding

Interactive machine learning from multiple biodata sources
Kaski, S., Reinvall, J., Chen, Y., Daee, P., Qin, X., Jälkö, J., Pesonen, H., Blomstedt, P., Eranti, P., Hegde, P., Siren, J., Peltola, T., Celikok, M. M., Sundin, I., Kangas, J., Afrabandpey, H., Honkamaa, J., Shen, Z. & Aushev, A.
01/01/2016 → 31/12/2018
Project: Academy of Finland: Other research funding

Computational models and methods for deciphering evolutionary patterns in bacterial genomic data
Marttinen, P., Kumar, Y. & Poyraz, O.
01/09/2015 → 31/08/2020
Project: Academy of Finland: Other research funding