Learning in natural and artificial systems

There are three possible options when facing an unknown situation. The first one is to exploit your "evolutionary memory"; you might not have seen this situation, this environment, before, but you did experience some others. So, throw the dice and wait to see if you get lucky, and one of your already-tested responses also works in this case. A second alternative is to apply your set of response systems. A kit of if-then routines that you acquired over decades of evolutionary vagaries. If these signals suit the response system, then trigger the response. We have studied several types of these routines in bacterial systems trying to understand how structure leads to function. Moreover, both of these situations are necessarily associated with your evolutionary past. But what about designing a "smart" system that works on the fly, a system that can adapt to everyday problems. How does this system evolve? How does it work? These questions take us to neurobiology and also to the design of artificial systems. Conscience? Umm.

Some readings

Deconstructing a multiple antibiotic resistance regulation through the quantification of its input function. With D. Bajic, I. Elola, and G. Rodrigo (2017). [PDF].

Genetic redundancies enhance information transfer in noisy regulatory circuits. With G. Rodrigo (2016) [HTML].

Antagonistic autoregulation speeds up a homogeneous response in Escherichia coli. With D. Bajic, I. Elola, and G. Rodrigo (2016). [PDF].

Trade-offs and noise tolerance in signal detection by genetic circuits. With J. Estrada, and R. Guantes (2010). [HTML].

Multistable decision switches for flexible control of epigenetic differentiation. With R. Guantes (2008). [HTML], [Sup].