June 4, 2019

Yann LeCun Delivers Master Class on AI and Identity at AFF Disrupt 2019

Element’s Chief Growth Officer and Managing Director for Africa, Barbara Iyayi, took to the main stage at the Africa Fintech Foundry (AFF) Disrupt 2019 Conference, sponsored by Access Bank, to introduce Yann LeCun - Chief AI Scientist at Facebook, professor at New York University, Turing Award Laureate, and Co-Founder of Element - who delivered a master class on AI and identity.

LeCun, recognized for his work on convolutional neural networks and his instrumental role in developing the modern field of Deep Learning, announced Element’s burgeoning partnership with Access Bank to use AI-powered identity verification and authentication to improve access to financial services in Nigeria. In particular, LeCun noted Element’s crucial focus on inclusivity to ensure access no matter who or where you are - evidenced by Element’s success in delivering identity across the planet with AI techniques built with everyone in mind.


“Element - here in Nigeria - is working with Access Bank, which is by far the largest bank in Nigeria, to use AI-based identity verification and authentication to help people create bank accounts and authenticate financial operations, so that financial services can be provided to them without the fear of their identity being stolen or faked.”


Beyond announcing Element and Access Bank’s partnership, LeCun explained why many economists view AI as a General Purpose Technology (GPT) - a technology that can affect an entire economy at the national and global level - detailing its transformative nature across a variety of sectors including healthcare, financial services, manufacturing, and transportation. Specifically, LeCun emphasized AI’s key role in revolutionizing communication and its importance in Africa.


“One very important aspect of AI today is that we have language translation systems that work quite well … this may allow people to communicate across language barriers. There are so many languages in Africa - basically where human language was born - and the diversity of languages is so great that it needs to be preserved. Allowing people to speak in their own language - to express themselves in their own language - and using translation to communicate with people across language barriers is crucial.”


Translation systems built with supervised Deep Learning techniques require an abundance of parallel data of texts. According to LeCun, we would have to train 49 million machines to translate every language into every other language - which we do not have enough data to accomplish. With this challenge in mind, LeCun highlighted self-supervised learning, a technique which would allow machines to learn the nature of and represent language without being trained to do a particular task. In this case, neural networks are able to produce the meaning of a sentence, regardless of the language. On top of this, additional neural networks can produce translations of that text in any language - training a system to develop a universal system of language.

LeCun explained how these modern AI techniques can aid those in countries where identifying themselves is a challenge. “Identity is key. It’s certainly a way to give more people more access to more services,” LeCun noted. “Using modern AI techniques for identity and authentication … brings the possibility of [many] new services, particularly financial services, but also health services … and government services as well.”

Emphasizing Element’s innovative Deep Learning approach to identity, LeCun avered the utility of these techniques for reliable identity verification, particularly in circumstances where the infrastructure for identity is not fully developed. In the case of facial recognition, less advanced systems require millions of images of individuals as well as a diverse set of images for each person, limiting a system to recognize only the individuals within the gallery that it has been trained on. Element’s platform goes beyond these limitations. “The system,” as LeCun describes, “Is universal - it can be applied to any face and authenticate any face. That’s why machine learning, AI, and Deep Learning are the key technologies for authentication and identification.”