2023 Deep Learning Indaba (DLI) Experience!

Rancy Chepchirchir
5 min readSep 28, 2023

--

Presenting my first poster on 'Generative style transfer for MR Image Segmentation: A case of the glioma segmentation in Sub-Saharan Africa’ at the DLI, 2023

Oh Yes, the DLI 2023 took place from the 03rd through 10th of September, 2023 at the University of Ghana, Ghana. I can now attest to all that couldn’t make it or have never attended it that this year, ‘The Deep Learning Indaba’ was more than just another conference; it was an epitome of inspiration and a rendezvous of some of the brightest minds in the African AI community. Reflecting on my experience a few weeks ago, the vibrant atmosphere, groundbreaking research, and the community’s camaraderie left an indelible mark on my professional and research journey.

The excitement started even before the event, knowing I’d reunite with colleagues from Strathmore University, The Sprint AI Training for African Medical Imaging Knowledge Translation (SPARK) Academy 2023 summer school on deep learning in medical imaging, and other brilliant minds I had made acquintance at the Big Data Summer School earlier this year. These prior connections only heightened my anticipation. The exchanges of ideas, the mutual respect, and the shared passion for data science created a palpable energy that permeated the entire venue.

Happy reunions at the Deep Learning Indaba (DLI), 2023

Presenting posters at such events is always a nerve-wracking experience. The fear of the unknown, the anticipation of feedback, and the hope for acknowledgment all mingle to form a cocktail of emotions. My first poster delved into the realm of medical imaging, titled “Generative Style Transfer for MR Image Segmentation: A Case of Glioma Segmentation in Sub-Saharan Africa”; a paper we worked on after taking part in the BraTS-Africa 2023 challenge whilst at The SPARK Academy. This was facilitated by the input of the amazing instructors of The SPARK Academy in providing insightful background knowledge on brain tumors that informed the research presented here; our amazing coordinator, country director, supervisor, and the incredible McMedHacks materials covering the Fundamentals of Medical Imaging. This research’s significance, especially for the African context, was well-received with both positive and constructive feedback. The validation came in the form of a beautifully crafted Ghanaian artifact, a recognition that will hold a special place both in my heart and my trophy cabinet.

Recipients of the Ghanaian Crafts at the Deep Learning Indaba (DLI), 2023

The second poster switched gears from medical imaging to Mathematical Finance, discussing the “Pricing of American Options: Numerical Approximations versus Physics-Informed Neural Network (PINN) Model” from my recently completed MSc thesis. The juxtaposition of traditional numerical methods with cutting-edge neural networks drew intriguing discussions and debates. Winning two signed copies of Kevin Murphy’s “Probabilistic Machine Learning” was the cherry on top. The book, a magnum opus in the field, was a perfect reward for the months of hard work.

Recipients of the 2 signed copies of Probabilistic ML by Kevin Murphy at the Deep Learning Indaba (DLI), 2023

Beyond the poster presentations, the InstaDeep hackathon on “Unveiling Cassava’s Secrets” hosted on Zindi was a whirlwind of coding, brainstorming, and innovative problem-solving for Accelerating African Agrigenomics. Cassava (Manihot esculenta) is the main food staple in Ghana, crucial for sustenance and income for many families and makes up 22% of Ghana’s Agricultural Gross Domestic Product (AGDP). Teaming up with diverse professionals, each bringing a unique perspective, epitomized the essence of collaborative learning. The tight deadlines and the race against time made the experience exhilarating.

Nonetheless, the workshops were perhaps the highlight of the Indaba. Each was a deep dive into niche areas of AI, expanding horizons and challenging preconceived notions. The “Weakly Supervised Computer Vision” workshop unveiled the vast potential of leveraging limited labeled data, a common challenge in real-world scenarios. “Data Science for Health in Africa” shed light on the myriad ways AI can revolutionize healthcare, especially in resource-constrained settings as seen in Sub-Saharan Africa. The “Ro’ya Computer Vision for Africa” workshop was particularly close to my heart, emphasizing the importance of contextual solutions tailored for the African continent. Lastly, “Building a Global Network of AI Researchers on AU and UN SDGs” was a clarion call for researchers to align their efforts with global goals, emphasizing the transformative potential of AI for societal good.

A standout addition to the program was the “Breakfast and Mentorship” sessions held throughout the Indaba week. The opportunity to interact and learn from senior researchers in the industry was invaluable. It wasn’t just a meal but a buffet of wisdom, insights, and guidance. Sitting across the table from Murphy, Miles, Cintas, Lee, to name a few, discussing AI’s nuances, challenges, and future directions was transformative. The informal setting allowed for candid conversations, bridging the gap between budding researchers like myself and the industry’s giants. It was mentorship in its truest sense, and the lessons learned during those breakfast sessions will be cherished for a lifetime.

After a 'Breakfast and Mentorship' session with Mr Lee of OpenAI at the DLI, 2023

In conclusion, the Deep Learning Indaba was not just an event; it was a celebration of knowledge, collaboration, and the shared vision of a brighter future powered by AI across the African continent. As I look back, the memories, the learnings, and the connections made will undoubtedly shape my professional journey in the years to come. The Indaba was a testament to the African AI community’s resilience, innovation, and immense potential. See y’all at the 2024 Deep Learning Indaba! Yebetumi!

--

--