• Reinforcement learning (AI)

    Microsoft Reactor

    Agenda: 6.00 - 6.30 Mingle and networking 6.30 - 7.30 Talk and questions 7.30 - 8.00 networking and wrapping up Talk title: Reinforcement learning in Julia Abstract: Joel will talk about his experiences doing reinforcement learning in Julia to play Atari games and get simulated robots to learn how to walk. Bio: Joel Mason TBD

    2
  • How to use Julia for optimisation and power systems research

    Abstract: Currently, power systems research in Julia focuses on static approximations to the problem of power flow in grids, i.e. obtaining voltage and current throughout grids given expected operation of generators and consumption patterns. PowerModels.jl is a toolbox that implements a variety of power flow models. It supports both simulation and optimisation of power grids. It is built upon JuMP.jl, a toolbox for mathematical optimisation for Julia. Today, JuMP is used around the world in the context of operations research. It allows the user to specify and solve optimisation problems without expert knowledge, yet at the same time allows experts to implement advanced algorithmic techniques. JuMP is also fast - benchmarking has shown that it can build models at speeds similar to commercial tools. The talk will introduce JuMP and PowerModels, and illustrate some of the problems that can be solved using these packages. Bio: Frederik Geth is a researcher with CSIRO Energy Newcastle. He is a Belgian, he is an electrical engineer and has been using Julia since v0.3. He performs research on the optimisation of power grids using Julia. His research interests include accurate (from the physics perspective) and scalable optimisation models for power systems. In this context, applications of such optimisation models include model predictive control, market clearing and reliability management. Agenda: 6.00 - 6.30 Mingle and networking 6.30 - 7.30 Talk and questions 7.30 - 8.00 networking and wrapping up

    1
  • Julia for rapid prototyping of realtime audio processing

    **Agenda** 5.45pm - 6.30pm --- Mingle and set-up 6.30pm - 7.15pm --- Talk: Julia for rapid prototyping of realtime audio processing 7:15pm - 7.30 --- Q&A --------------------------------------------------------------------------------------------- Speaker: Rowan Katekar Bio + Abstract: Since 2014 I have been working at Dolby in Sydney as a software engineer. I spent 2 years working in a product team on a core C signal processing library, before moving into research. I now work as part of the Speech Analytics team, where our focus is identity and intent in the context of distributed and asynchronous audio. Earlier this year, with support from Dolby, I started as a part-time PhD student at UNSW in this research area. Moving to Julia from MATLAB + C has drastically cut down the time it takes to go from idea to implementation. I'll be bringing along some live demos of the capabilities we have developed using Julia.

    1
  • Why I Chose Julia, or, an Exercise in the Statistical Bootstrap

    **Agenda** 5.45pm - 6.30pm --- Mingle and set-up 6.30pm - 7.00pm --- Talk: Why I Chose Julia, or, an Exercise in the Statistical Bootstrap 7.00pm - 7.30pm --- Talk: Fast Group By and String sorting in Julia --------------------------------------------------------------------------------------------- Title: Why I Chose Julia, or, an Exercise in the Statistical Bootstrap Abstract: In this presentation, I discuss the reasons that in 2014, I chose to convert my entire workflow to Julia. The reasons are explained using my registered Julia package, DependentBootstrap, as a contextual example. Author: Colin T. Bowers Bio: I am a time-series econometrician, specialising in financial econometrics. I completed a doctorate at Macquarie University in 2015, titled "Estimation and Forecast Evaluation of Risk Measures with High Frequency Financial Data". I currently maintain a small workload at Macquarie University, and spend most of my time working on forecasting a range of financial random variables.

    1
  • Why I Chose Julia, or, an Exercise in the Statistical Bootstrap

    **Agenda** 5.45pm - 6.30pm --- Mingle and set-up 6.30pm - 7.00pm --- Talk: Why I Chose Julia, or, an Exercise in the Statistical Bootstrap 7.00pm - 7.30pm --- Talk: Fast Group By and String sorting in Julia --------------------------------------------------------------------------------------------- Title: Why I Chose Julia, or, an Exercise in the Statistical Bootstrap Abstract: In this presentation, I discuss the reasons that in 2014, I chose to convert my entire workflow to Julia. The reasons are explained using my registered Julia package, DependentBootstrap, as a contextual example. Author: Colin T. Bowers Bio: I am a time-series econometrician, specialising in financial econometrics. I completed a doctorate at Macquarie University in 2015, titled "Estimation and Forecast Evaluation of Risk Measures with High Frequency Financial Data". I currently maintain a small workload at Macquarie University, and spend most of my time working on forecasting a range of financial random variables.

    1
  • Intro to Julia from a Data Perspective - Can it be the fastest too?

    Feel free to come early around 5.45pm to mingle and watch a Julia video. Talk starts at 6.30pm Intro to Julia from a Data perspective - Can Julia be the fastest in-memory data manipulation language too? Presenter: Dai ZJ (http://www.linkedin.com/in/daizj) Bio: ZJ has worked in banking & data science related fields since 2007. He is a part of a startup in stealth where being able to manipulate 100GB size datasets reasonably quickly without expensive hardware/software is important. Abstract: We all know Julia is meant to be fast. But there is an important caveat - it is fast at compute tasks but is it fast at data tasks where R's data.table is king? Can it be faster than data.table? This talk will offer an introduction to Julia's data ecosystem and gives an "into the future" look at what's around the corner for Julia as a data manipulation tool.

  • Numerical computing with functions in Julia

    Microsoft Store

    Numerical computing with functions in Julia Presenter: John Wormell (http://www.maths.usyd.edu.au/u/wormellj/) Bio: PhD student at the University of Sydney interested in numerical methods for dynamical systems, chaos, spectral bases Abstract: While all we classically expect of a function in programming is that it produces outputs from inputs, knowledge of the function as a whole is really required for many operations: these include calculus operations, differential equation solving, root-finding and sampling from a distribution. This talk will introduce some Julia packages that implement various kinds of natural, highly efficient "whole-function" manipulation. It will focus on ForwardDiff.jl, which performs efficient automatic differentiation, and ApproxFun, which represents functions very accurately as a sum of Chebyshev polynomials and then solves many problems using numerical linear algebra. We will discuss in particular how these packages make use of Julia's capabilities, including multiple dispatch and the parametric type system.

    5
  • Introduction to Julia

    Microsoft Store

    Introduction to Julia Presenter: Joel Mason Bio: PhD student at UNSW interested in Machine/Deep/Reinforcement Learning, Interactive Data Visualisation, AI, and maintainer/contributor to various Julia packages as https://github.com/JobJob Abstract: The Julia programming language ( https://julialang.org/ ) is a high level language specialised for numerical computing, originally developed at MIT, that was first publicly released in 2012. Julia code looks a bit like Python or Matlab, but is designed from the ground up with a focus on high performance (approaching that of C). This talk will give an introduction to Julia, and discuss features of the language that make it well suited for machine learning, data science, and scientific computing in general. It will also discuss the challenges and opportunities of working with a relatively young language, with a smaller software library ecosystem/open source community than the likes of R, Python/NumPy/SciPy, and Matlab.

    2
  • First Julia meetup hosted by Microsoft

    Microsoft Store

    Seeking speakers to present 15-40mins! Please message me if you want to speak. First ever session of Julia meet up group! Please share with fellow Julia users!