Add this schedule to your google calendar using this link.
Indaba 2018: Schedule
-
Sunday, 9 Sep
Welcome Day and Background
-
-
09:00 - 18:00
StellenboschArrivals in Stellenbosch and check-in
-
11:00 - 14:00
Van der Sterr FoyerRegistration
-
14:00 - 16:00
Computer LabsIn this practical, we introduce the idea of classification using a machine-learning model. We explore the relationship between a classifier’s parameters and the decision boundary and also introduce the idea of a loss function. Finally, we briefly introduce TensorFlow. This session will also be a great opportunity to meet your tutors and fellow attendees!Stephan Gouws; Avishkar Bhoopchand; Ulrich Paquet, DeepMind -
16:30 - 18:30
Van der Sterr BuildingLearn the key concepts of probability and differential calculus necessary to gain the most from the rest of the week’s programme. This session will be split into 5 smaller classes run in parallel (each covering the same material).Avishkar Bhoopchand; Cynthia Mulenga; Daniela Massiceti; Kathleen Siminyu; Kendi Muchungi -
18:30 - 19:00
EndlerTransfer to Root44: Coaches will depart from Endler Hall to Root44. Last bus will depart at 19:00 sharp. Busses will return from Root44 to Endler Hall between 21:00 and 23:00. Parking is available at Root44 for delegates who will travel in own cars. -
19:00 -
Root44Experience a warm Winelands welcome at our Opening Event on Sunday, 9 September! Our venue is Root44, famous for good food and local music. Root44 is located on a historic wine farm and surrounded by gorgeous vineyards and trees. Guests will be treated with gourmet hamburgers, wood fired pizzas and decadent sweet treats paired with a selection of local beer, award-winning wines from the region and soft drinks.
-
-
Monday, 10 Sep
Fundamentals and Convolutional Models
-
-
07:30 - 08:30
Endler FoyerRegistration and Morning Coffee
-
08:30 - 09:30
Endler Hall -
09:30 - 10:30
Endler Hall -
10:30 - 11:00
Endler FoyerBreak
-
11:00 - 12:00
Endler Hall -
12:00 - 14:00
Neelsie & Endler FoyerLunch and Poster Session
-
14:00 - 16:00
Computer LabsIn this practical we implement and train a feed-forward neural network (also know as an “MLP”) on the famous MNIST dataset of handwritten digits. We consider the practical issues around generalisation to out-of-sample data and introduce some important techniques for addressing this.Stephan Gouws; Avishkar Bhoopchand; Ulrich Paquet, DeepMind -
16:00 - 16:30
Endler FoyerBreak
-
16:30 - 18:30
Endler Hall -
18:30 - 19:30
Free Time
-
19:30 - 21:00
Pulp CinemaAlphaGo Movie Screening
200 people maximum. Sign up at registration desks -
19:00 - 22:00
STIASWomen in Machine Learning
Generously hosted by Microsoft
-
-
Tuesday, 11 Sep
Recurrent Models
-
-
07:30 - 08:30
Endler FoyerMorning Coffee
-
08:30 - 10:30
Computer LabsIn this practical we will cover the basics of convolutional neural networks or “ConvNets”. This type of architecture has had tremendous success in both vision and sequence modelling. We cover some of the important details around the convolution operation before delegating the heavy-lifting to TensorFlow to build a model that to classifiy CIFAR10 images.Stephan Gouws; Avishkar Bhoopchand; Ulrich Paquet, DeepMind -
10:30 - 11:00
Endler FoyerBreak
-
11:00 - 12:00
Endler HallThis session aims to build intuitions around probability spaces before covering sampling, bayesian methods and briefly introducing variational inference for quantifying uncertainty in deep learning.Yabebal Fantaye, AIMS South Africa -
12:00 - 14:00
Neelsie & Endler FoyerLunch and Poster Session
-
14:00 - 16:00
Endler Hall -
16:00 - 16:30
Endler FoyerBreak
-
16:30 - 18:30
Computer LabsIn this practical we move from feed-forward models to sequence models which are designed to model the dependencies between inputs in sequences of data that change over time. We consider the fun problem of language modelling of Shakespeare text and build a model that can generate realistic-looking text in the style of Shakespeare.Stephan Gouws; Avishkar Bhoopchand; Ulrich Paquet, DeepMind -
18:30 - 19:00
Free Time
-
19:30 - 21:00
Pulp CinemaAlphaGo Movie Screening
200 people maximum. Sign up at registration desks -
19:30 - 21:30
Van der Sterr, theatre 1046Research must be communicated in order to have impact, and the primary mode of communication is through writing research papers. This session approaches the writing process from the angle of the researcher (improving your own understanding through writing), the reviewer (avoiding common pitfalls that get papers rejected), and the reader (bringing your message across and making your work as impactful as possible).Nando de Freitas, Ulrich Paquet and Stephan Gouws, DeepMind; Martin Arjovsky and Kyunghyun Cho, New York University
-
-
Wednesday, 12 Sep
Generative Models and Reinforcement Learning
-
-
07:30 - 08:30
Endler FoyerMorning Coffee
-
08:30 - 10:30
Endler Hall -
10:30 - 11:00
Endler FoyerBreak
-
11:00 - 12:00
Endler HallKambule and Maathai Awards session
-
12:00 - 14:00
Neelsie & Endler FoyerLunch and Poster Session
-
14:00 - 16:00
Endler Hall -
16:00 - 16:30
Endler FoyerBreak
-
16:30 - 18:30
Computer LabsIn this practical we introduce the idea of reinforcement learning, discuss how it differs from supervised and unsupervised learning and then build an agent that learns to play a (simple) game!Stephan Gouws; Avishkar Bhoopchand; Ulrich Paquet, DeepMind -
18:30 - 19:00
Free Time
-
19:30 - 21:00
Pulp CinemaAlphaGo Movie Screening
200 people maximum. Sign up at registration desks -
19:00 -
Revealed on registrationMentorship Evening Events (x2)
Two events, hosted by Capitec and Nedbank
100-120 maximum for each event. Sign up online or at sponsor booths
-
-
Thursday, 13 Sep
Parallel Tracks
-
-
07:30 - 08:30
Endler FoyerMorning Coffee
-
08:30 - 10:30
Van der SterrThis deep dive explores the exciting new progress in the last few years on sequence models which do not make use of recurrent neural networks. We will look at ways of modeling sequences using convolutions as well as so-called “attention models” such as the Transformer.Kyunghyun Cho and others -
08:30 - 10:30
Van der SterrLearned the basics of Convolutional Neural Networks? Want to know go beyond? Join to extend your understanding of CNNs and how they extract image features for higher-level computer vision tasks like object detection, localisation and semantic segmentation. Following this, we invite a panel of computer vision experts to give their personal insights, advice and expert views on the frontiers of the field: what are the biggest unsolved problems in computer vision, how are they relevant to Africa, and where should African researchers be directing their energy to solve these problems. Also join to hear short spotlight talks given by fellow Indaba attendees - a great opportunity to learn more about current state-of-the-art methods being used in computer vision!Daniela Massiceti, University of Oxford; Sara Hooker, Google Brain, Saumya Jetley, University of Oxford
MORE INFORMATION HERE -
08:30 - 10:30
Van der SterrIn this session we wish to highlight the ongoing as well as new directions of AI/ML work happening in Africa. This session will have two sections, first we will have short presentations 5-10 minutes from senior representatives from AI/ML/Policy companies and organisation (both NGO and governmental). These presentations are meant to provide an overview of what each organisation offers, highlight their successes, learn from challenges encountered and understand opportunities discovered. Following the presentations, we will have a Q&A session to open up the conversation to the audience.Sumir Panji, Linet Kwamboka, David Sengeh, Jon Lenchner, Moustapha Cisse -
10:30 - 11:00
Endler FoyerBreak
-
11:00 - 12:00
Endler HallBeginning with the subject of “Strengthening African Machine Learning”, this fireside chat shall be an opportunity for the diverse African and International audience to converse, share experiences, advice, strategies and challenges around the current state of ML on the continent and what the future may hold. -
12:00 - 14:00
Neelsie & Endler FoyerLunch and Poster Session
-
14:00 - 16:00
Endler Hall -
16:00 - 16:30
Endler FoyerBreak
-
16:30 - 18:30
Van der SterrThis sessions builds on the understanding of generative models, with a focus on applications in healthcare. For the first half, Konstantina will discuss the role of probabilistic thinking, uncertainty and causality, and then look at how these tools can be used to build personalised healthcare tools. In the second part, Shakir will recap the area of generative models, specifically the algorithms for LDA, VAEs and GANs, and then look at how these can be applied in healthcare settings ranging from analysis of electronic health records, medical notes, in drug discovery, and in medical imaging.Konstantina Palla, Microsoft Research; Shakir Mohamed, DeepMind -
16:30 - 18:30
Van der SterrLearn about the recent history of NLP and discuss the biggest open problems in NLP with a panel of experts.Herman Kamper, University of Stellenbosch; Sebastian Ruder, AYLIEN -
16:30 - 18:30
Van der SterrAre you itching to innovate? Are you dreaming about changing the future? Would you like to learn about VC funding? About starting tech companies and growing them? This is the session for you!Andrea Böhmert, Knife Capital; Karim Beguir, InstaDeep; Blaise Thomson, Apple -
18:30 - 19:00
Transfer to Evening Event
-
19:00 -
Seven SistersDrawing inspiration from the roots and heritage of the seven sisters (and one brother) who own the farm, our Closing event will offer a taste of the Cape. We're talking an authentic, vibrant and wholesome braai (barbeque) served with delicious salads, breads, preserves and traditional desserts. The ever so famous DJ Renee will provide fabulous dance music to end the Deep Learning Indaba on a high note!
-
-
Friday, 14 Sep
Closing
-
-
07:30 - 08:30
Endler FoyerMorning Coffee
-
08:30 - 10:30
Van der SterrParallel: Reinforcement Learning II
Talk: Benjamin Rosman, CSIR. Panel: David Silver, DeepMind; Katja Hofmann, Microsoft Research Cambridge -
08:30 - 10:30
Van der SterrLearn the tricks of the trade for deploying and scaling ML models in a production environment from experienced practitioners. This session will cover topics ranging from scaling machine learning systems to a deep-dive into the practicalities of building APIs around ML models.Stuart Reid, NMRQL Research; Amine Kerkerni, InstaDeep; Omoju Miller, GitHub -
08:30 - 10:30
Van der SterrThis session will tackle the intersections of AI/ML, Ethics and Policy on the continent. Session will be a blend of a practical interactive Ethics session, a talk on fairness and robust discussion via an expert panel made up of researchers, practitioners, policy makers. At the end of the day, we would like to answer: How do we work to inject our own values into AI/ML development in Africa, allow a progressive environment for development and protect our communities?Facilitators: Vukosi Marivate; Muthoni Wanyoike
Speakers: Timnit Gebru; Mmaki Jantjies; Osonde Osoba; Linet Kwamboka -
10:30 - 11:00
Endler FoyerBreak
-
11:00 - 13:00
Endler Hall -
13:00 - 13:30
Endler HallClosing, Poster Prizes, Farewell
-
13:30 - 14:30
NeelsieLunch
-