Deep Learning Computer Vision Machine Learning Artificial Intelligence Data Science
Melbourne, Victoria, Australia
Passionate in technology innovation to make people's life easier and with over 25 years of experience in software development, Gus has been working in various industries from 3D/Animation, Games Development, Desktop Software, mobile apps and very recently Computer Vision, Machine Learning and Artificial Intelligence. Well-known for unorthodox ways of solving difficult problems, Gus currently works at Carsales as the Head of AI, inventing and building many cool AI projects such as Cyclops Image Recognition and providing technical direction to various teams across Carsales on AI.
Gus has an extensive experience in Deep Learning, Convolution Neural Network, Tensorflow, Keras and also loves speaking on AI and future tech related topics at various AI conferences.
Passionate in technology innovation to make people's life easier and with over 20 years of experience in software development, Gus has been working in various industries from 3D/Animation, Games Development, Desktop Software, mobile apps and very recently Computer Vision and Artificial Intelligence. Well-known for unorthodox ways of solving difficult problems, Gus is currently the Head of AI at Carsales, inventing and building many cool AI projects such as Cyclops Image Recognition.
How close are we to an AI Singularity? As many have also call it an Artificial General Intelligence (AGI), it is unlike the "narrow" AI we currently have today which can easily beat us at one specific task but fail to perform a simple task which requires combination of knowledge from multiple domains.
Most technical explanation in this talk will be done at intuition level to make sure that anyone at any technical level can enjoy and understand.
I will start with the introduction of what Artificial Intelligence is, followed by a comparison how we solve problem the non-AI vs AI way.
Furthermore, I will explain how AI can magically solves any problems as long as you can describe your problem mathematically. Next, I will show you two variations of AI, Machine Learning and Deep Learning and what are the difference between them.
I will then continue to explain our current progress in Deep Learning - starting with image recognition which is the most successful use of Deep Learning so far. How we use Convolution Neural Network (CNN) to recognise images which mimics human visual cortex. I will show you a demo of what's the state of the art CNN model can do today and how we express train them with transfer learning and distributed training. CNN is awesome but there are some limitations which soon may be addressed by latest research finding.
Next, I will tell you the latest trend in generative AI which can mimic human creativity through technique such as Variational Auto-encoder and GAN. I will also show you how we use them to build something like what we only normally see in science fiction movie such as image content reconstruction, mind reading and super resolution.
In NLP space, also equally interesting we can make AI understand linguistic with technique like Word embedding. I will show you what we can use it for. Furthermore, by combining it with LSTM we can give ability to AI to even write a poem.
Then, I will show you why some tasks are so easy for us human to do but not for AI even using the latest state of the art technology. Why it is hard to build an AGI with the lack of better knowledge representation and knowledge transfer across totally different domain. Which is why we need a totally new approach and why we need a centralised/cloud brain model.
Join Carl and Richard from .NET Rocks as they discuss artificial intelligence with Agustinus Nalwan.
In this session Agustinus Nalwan takes questions from Carl Franklin, Richard Campbell and the audience.
16 Sep 2018 - 20 Sep 2018
Sydney, New South Wales, Australia