Past Meetup

Sydney Machine Learning, Thurs, 22nd of Feb @ Domain.com.au

This Meetup is past

214 people went

Location image of event venue

What we'll do

• What we'll do
PLEASE NOTE: that RSVPing to this page DOES NOT GRANT YOU ACCESS to this meetup,
Spaces are limited! Due to popular demand & Domain Facilities Management, PLEASE SIGNUP via the EVENTBRITE LINK BELOW:

>>CLICK EVENTBRITE LINK BELOW<< -

https://www.eventbrite.com.au/e/sydney-machine-learning-thurs-22nd-of-feb-domaincomau-tickets-42932687825 -

PLEASE NOTE: that RSVPing to this page DOES NOT GRANT YOU ACCESS to this meetup, please signup via the eventbrite link ABOVE...
_________________________________________

February is an absolutely crazy month for Sydney Machine Learning! With two meetups & the kickoff of our FasterAi Study Group!

In another first, Domain.com.au has been kind enough to be our host for this upcoming Sydney Machine Learning meetup. Come along for beer, pizza, networking and talks from Benjamin Wilson PhD, Chief Scientist at Lateral - who will be talking about Simulating Turing Machines with Recurrent Neural Nets and Rinat Sadykov CTO of Cozitrip who will be talking about how to implement neural networks for facial recognition.

First Presentation Summary:

Facial recognition - Ever wondered how the neural nets used for everything from face payments to tracking of criminals work?

Rinat will go over how neural networks are used for everything from the detection of a face to identifying the person, as well as an overview of what the deep learning network is doing in the different stages.

Rinat was a software engineer at Atlassian. Rinat is now the Chief Technical Officer of Cozitrip.

Second Presentation Summary:

“Simulating Turing Machines with Recurrent Neural Nets”

Proponents of recurrent neural nets (RNNs) can often be heard boasting that "RNNs are Turing complete". That is, they claim that for any computable function, there is a (carefully constructed) RNN that computes this function.

This means that there is even an RNN that simulates a universal Turing machine -- so there is a single RNN that can compute any computable function!

In this talk, Ben will present the highlights of the now classical (1995) paper of Siegelmann and Sontag, where this extraordinary statement is proven. As we'll see, the key to the proof is a beautiful construction for encoding the discrete, symbolic machinery of the Turing machine (stacks holding bits and stack operations like pop, push, etc.) in the continuous, rational-valued world of recurrent neural nets.

Ben leads the research of Berlin-based startup Lateral GmbH, working primarily on problems involving natural language and user behaviour.

Time: 6:00 – 8:30 pm

Venue: Domain, 55 Pyrmont St, Pyrmont, Sydney

Schedule:

6:00 - 6:30: Beer, Pizza, Networking

6:30 - 6:40: Intro

6:40 - 7:00 Rinat’s Talk

7:00 - 7:40 Ben's Talk

7:40 - 8:00 Wrap up

_______________________________________

• What to bring
have you signed up via the eventbrite link on this page?

• Important to know
have you signed up via the eventbrite link on this page?