Google’s Compute for Underrepresented Researcher’s programme (CURe) has partnered with the Deep Learning Indaba’s Mentorship Programme to host an outreach event at this year’s Deep Learning Indaba in Accra, Ghana. The outreach event will be a series of short call-for-collaboration presentations, presented in a specific format called Requests for Plots (RFPs).

What is the Request for Plot format?

The Request for Plot format was popularised by ML Collective and is aimed at kickstarting collaborations. An 80-second introduction of this format can be viewed below:

What will happen at the Deep Learning Indaba?

There will be a planned track at the Indaba where attendees can listen to these pitches brought to the table by Google researchers. Google Researchers will describe the problem, give a concrete description of the first plot that one would produce to start the project. Attendees of the Indaba will have the opportunity to listen to these pitches during a session at the Indaba. There will be an opportunity to ask questions and then it will be up to those interested to work towards that first plot.

Once you have completed the plot there will be a link for you to submit your work – then, if you are selected you will continue working on the problem with a Google researcher through a collaborative effort.  If selected, GCP credits to help support the research will be provided.

What kind of topics can you expect?

There will be a wide range of topics presented at the sessions. Browse the list of research titles below to get a sense of what you might be interested in:

  • Online Bayesian Neural Networks with GP-inspired Priors for Spatio-Temporal Forecasting
  • Neural network with smart allocation of compute
  • MazahaGPT — Building a ChatGPT like model for the African Subcontinent
  • Developmental and validation study on wearables
  • LLMs for Under-Represented Languages
  • Benchmarks for African Cultural Erasure in Large Language Technologies 
  • duellm: Adversarial Attacks on Generative Models using other Generative Models
  • AI and Climate Information in Africa
  • Black Language Inclusion in NLP
  • Efficient Synthesis of Training Datasets From Black-Box Generative Models

What do you need to do?

Before the presentations on September 7, you are welcome to register your interest and ask clarifying questions here. During the presentations, be ready to listen and to engage with the speakers by asking questions to get further details. After the Indaba you’ll want to work towards the problem that interests you most.

When will the session be hosted?

The session is scheduled for Thursday, September 7th from 16:30 to 18:00. Be sure to consult the Deep Learning Indaba programme for the most updated information and locations during the conference!

When will the plots be due?

The form to submit the plots will close 6 months after the conference (March 5, 2024).  However, the sooner you submit your plot, the more likely it will be to be selected for collaboration with a Google researcher!