#38 - 2021.05 - GrokNet and NLP for low-resource languages

And we are back with more interesting talks and amazing speakers! :)

🆂🅻🅾🆃 #0: 18:00-18:10 - Welcome

🆂🅻🅾🆃 #1: - 18:10-18:55 - GrokNet – Unified Computer Vision Model Trunk and Embeddings for Commerce, Pratik Dubal, Facebook AI

GrokNet is a deployed image recognition system for commerce applications. GrokNet leverages a multi-task learning approach to train a single computer vision trunk. It can identify fine-grained product attributes across billions of photos — in different categories, such as fashion, auto, and home decor. It achieves a 2.1x improvement in exact product match accuracy when compared to the previous state-of-the-art Facebook product recognition system.

👨🏻‍🔬 Pratik Dubal is a Software Engineer at Facebook AI working on Computer Vision for Commerce. Prior to joining Facebook, Pratik completed his MS in Computer Science from Columbia University. He previously led the ML team at AitoeLabs, a video analytics start-up, where he worked on object detection, fine-grained classification, and video summarization.

🆂🅻🅾🆃 #2: - 19:00-19:45 - NLP advances for African Languages - Bonaventure Dossou + Chris Emezue

Language is inherent and compulsory for human communication. Whether expressed written, spoken, or signed, language ensures understanding between people of the same and different regions. African languages (over 2000) are complex and truly low-resourced. These languages receive minimal attention: the datasets required for NLP applications are difficult to discover and existing research is hard to reproduce. However, this is changing. With the growing awareness and effort to include more low-resourced languages in NLP research, African languages have recently been a major subject of research in natural language processing. The presentation is the story of the dynamic duo, Chris and Bonaventure, as they work towards tackling the NLP challenges facing African languages. The talk will cover a range of topics with a focus on the OkwuGbé end-to-end speech recognition system for Fon and Igbo.

👨🏿‍🔬 Bonaventure Dossou: Bonaventure Dossou is a Masters's Student at Jacobs University, Bremen, and a research Intern at Mila Quebec AI Institute, working under the supervision of Professor Yoshua Bengio. His research areas include Machine & Deep learning (and its application in computer vision, natural language processing for Healthcare and African Languages), Computational biology, Network Science Approaches in Biology and Medicine.

👨🏿‍🔬 Chris Emezue: Chris Emezue is a master's student at the Technical University of Munich. He works with SIEMENS and also recently joined the Mila Institute as a research intern. His research areas include machine/deep learning, reinforcement learning, multilingualism, knowledge-graph representations, speech processing, and AfricaNLP.

📌 Do not miss out on updates on Twitter - https://twitter.com/hack_ai and join the Munich ML community on Slack channel https://ai-hack-inviter.herokuapp.com/