19.05.2014

David Janssen's new abstract accepted as a poster at ECVP from 24.08.-28.08.2014

Title: "Improving models of early vision through Bayesian analysis" by David Janssen and Felix Wichmann

Abstract:

Improving models of early vision through Bayesian analysis

Computational models are often used in vision science to formalize theories about how the visual system functions. The most general models attempt to relate visual input---images---to results obtained through psychophysical experiments.
Typically, image-based models contain many parameters. Maximum likelihood estimation (MLE) is a commonly used technique to statistically estimate the model parameters from the data, but MLE only provides point estimates. Even when MLE is complemented by a variability estimate based on the normal approximation at the posterior mode, it does not explore the confidence regions of, and dependencies between, model parameters in sufficient detail.
We present an image-driven model of early vision within a Bayesian framework in which we estimate the posterior distributions of the parameters via Markov Chain Monte Carlo (MCMC) sampling. This provides us both with a "best fitting model" and estimates of the confidence intervals on and correlations between the model parameters. We demonstrate how this information helped guide our model-design and avoid non-obvious pitfalls that would not have been apparent via MLE.
Finally, we provide an aside on why subband decompositions do not function well within standard models of early vision and provide a computationally intensive but functional alternative.