Personal tools
You are here: Home Events ANC Workshop - Maciej Pajak/Amos Storkey Chairing: Jim Bednar

ANC Workshop - Maciej Pajak/Amos Storkey Chairing: Jim Bednar

— filed under:

  • ANC Workshop Talk
When Jan 13, 2015
from 11:00 AM to 12:00 PM
Where Room 4.31/4.33
Add event to calendar vCal

Maciej Pajak

Title:  Computational phylogenetics methods for uncovering evolution of synaptic function


Recent studies suggest that the evolution of the synapse proteome (from early protosynaptic protein complexes in unicellular eukaryotes to sophisticated machineries comprising of thousands of proteins) is responsible for the emergence of complex nervous systems and complex behaviour, however, level of analysis remains superficial.
Evolution of the phenotype of organisms occurs due to underlying evolution of their genotype on molecular level, studying phylogeny and selection pressure of a protein can provide valuable information not only about the function and origin of that protein as a whole but also about the contribution of specific sites to its current role in the organism.
In this presentation I will discuss methods for in-depth molecular phylogenetic analysis allowing for insightful observations at single codon resolution. I will expand on specific problems encountered at different stages of the analysis and propose how they can be solved in the context of synaptic proteins.


Amos Storkey     (research work done in collaboration with Christopher Clark)

Title:   Teaching Deep Convolutional Neural Networks to Play Go


Mastering the game of Go has remained a long standing challenge to the field of AI. Modern computer Go systems rely on processing millions of possible future positions to play well, but intuitively a stronger and more 'humanlike' way to play the game would be to rely on pattern recognition abilities rather then brute force computation. Following this sentiment, we train deep convolutional neural networks to play Go by training them to predict the moves made by expert Go players. To solve this problem we introduce a number of novel techniques, including a method of tying weights in the network to 'hard code' symmetries that are expect to exist in the target function, and demonstrate in an ablation study they considerably improve performance. Our final networks are able to achieve move prediction accuracies of 41.1% and 44.4% on two different Go datasets, surpassing previous state of the art on this task by significant margins. Additionally, while previous move prediction programs have not yielded strong Go playing programs, we show that the networks trained in this work acquired high levels of skill. Our convolutional neural networks can consistently defeat the well known Go program GNU Go, indicating it is state of the art among programs that do not use Monte Carlo Tree Search. It is also able to win some games against state of the art Go playing program Fuego while using a fraction of the play time. This success at playing Go indicates high level principles of the game were learned.