This meetup is for everyone in the field of Timeseries Forecasting and Event Analytics. We regularly present you with top tier speakers and fantastic networking opportunities. Our talks are technical and usually show new ways of getting better results, cost/time savings and new model classes. If you like learning (or teaching) innovative methods in our field, then you are one of us - see you soon!
Come to our next event, up-skill, meet like-minded people, talk shop, have fun. Showcase your work and be recognised for your achievements, feel the high of a well-received talk.
The Timeseries Talks are Back! We are very excited to have David Rawlinson and Gideon Kowadlo from Project AGI present a cutting edge RNN talk.
David and Gideon present a recurrent neural network memory that uses sparse coding to create a combinatoric encoding of sequential inputs. Using several examples, they show that the network can associate distant causes and effects in a discrete stochastic process, predict partially-observable higher-order sequences, and enable a DQN agent to navigate a maze by giving it memory. The network uses only biologically-plausible, local and immediate credit assignment. Memory requirements are typically one order of magnitude less than existing LSTM, GRU and autoregressive feed-forward sequence learning models. The most significant limitation of the memory is generalization to unseen input sequences. They explore this limitation by measuring next-word prediction perplexity on the Penn Treebank dataset.
About the speakers:
David has been working in applied AI/ML for approximately 20 years in both industry and academia. These include finance & retail business analytics, security / surveillance, and med-tech particularly genomics, bioinformatics and medical imaging. Many of these systems are used everyday in the real world from banking and parking enforcement to radiology and monitoring of financial transactions in many countries around the world. He studied computer science & artificial intelligence at Sussex University (UK) back at the end of the AI winter, later moving to Monash University in Australia to complete a PhD in computer vision for mobile robots. Afterwards he worked for Melbourne University and Data61 (formerly NICTA) for several years. David now works on research towards more general purpose machine learning looking at alternative methods of credit assignment and continuous learning.
Gideon has spent 20+ years bringing emerging technologies to life in Australia and overseas, in areas from Robotics to Human Motor Control and from Mobile to Bio-inspired AI. He gets directly involved as well as leading and empowering teams to make an impact. He is a co-founder of Outware Mobile, Hoist AI and Project AGI. Outware Mobile began in 2009 and quickly became Australia’s leading app developer before becoming part of Arq Group a few years ago. Hoist AI supports AI startups that want to improve the world.
How to find us:
Once you enter the Outcome.life building take the lift or the stairs to the top floor.
Our venue, Outcome.life will be asking all attendees to register for security reasons.
We'll start the talk at 6:10,pm come at 6:00pm for networking. Stay after the talk for networking as well.