Normative theories and statistical inference provide complementary approaches for the study of biological systems. A normative theory postulates that organisms have adapted to efficiently solve essential tasks and proceeds to mathematically work out testable consequences of such optimality; parameters that maximize the hypothesized organismal function can be derived ab initio, without reference to experimental data. In contrast, statistical inference focuses on the efficient utilization of data to learn model parameters, without reference to any a priori notion of biological function. Traditionally, these two approaches were developed independently and applied separately. Here, we unify them in a coherent Bayesian framework that embeds a normative theory into a family of maximum-entropy ‘‘optimization priors.’’ This family defines a smooth interpolation between a data-rich inference regime and a data-limited prediction regime. Using simple examples from neuroscience and gene regulation, we demonstrate that our framework allows one to address fundamental challenges relating to inference in high-dimensional, biological problems.
Bio
Gašper Tkačik joined IST Austria in 2011. Previously he was a postdoc with Vijay Balasubramanian and Phil Nelson at University of Pennsylvania, working on the theory of neural coding and specifically exploring population coding and adaptation in the retina. He finished my PhD at Princeton with Bill Bialek and Curt Callan in 2007, studying how biological networks can reliably transmit and process information in the presence of intrinsic noise and corrupted signals.He is broadly interested in uncovering general principles that underlie efficient biological computation.
If you would like to attend the talk, please sign-up via office@csh.ac.at
Gašper Tkačik will give a talk on Friday, February 17, 2023 at 2 PM in the Salon.