Anima Anandkumar
JupyterCon 2020 keynote speaker announcement
Anima Anandkumar is Bren Professor of Computing and Mathematical Sciences at Caltech, and the Director of Machine Learning at NVIDIA. She was the youngest named chair professor at Caltech, has received many awards in a short time span, and has been featured in documentaries and media outlets like PBS, Wired, Forbes, MIT Tech Review and others.
Beyond her star power, Anima is a relentlessly outspoken and influential woman of the times. She has been public about her #MeToo experience at a previous place of work, and she spearheaded the #ProtestNIPS campaign to re-brand the Neural Information Processing Systems conference (now known as NeurIPS). The latter earned her a recognition in the 2018 Good Tech Awards of The New York Times.
She also advocates for responsible use of artificial intelligence, taking a stand specifically against face recognition services sold to law enforcement. This technology is currently still sufficiently inaccurate to be unsafe for use in high-consequence applications, and the product that Amazon made available in 2018, in particular, had high error rates when classifying the gender of dark-skinned women. In an open letter to Amazon, more than two dozen scientists (including Anima) refuted the claims Amazon was making on the scientific evaluation of the technology. Now, the letter has many more signatures, and the Association for Computing Machinery has called for a moratorium on the use of face recognition in high-risk settings.
Anima has a zeal for expanding the reach and understanding of AI in society. With Yisong Yue, she founded in 2018 the AI4science initiative at Caltech, which connects experts in data science and AI with researches across disciplines of science and engineering. She conveys that zeal to the boards of non-profits like #GoBeyondResumes, helping tech with skills-based recruitment, and Behind her Eyes, using virtual reality to combat bias and discrimination.
Nothing less than a trailblazer, her most celebrated contribution to machine learning was devising ways to use tensor algorithms in building statistical models, and having the insight that tensors are the natural mathematical representation for multi-modal data. Nowadays, tensors are everywhere in AI, but in 2014, when she published her most-cited paper on tensor decompositions, they were still obscure. Deep learning was just taking off then!
She collaborated with Jean Kossaifi and others on the release of TensorLy, which provides software support for tensor operations in Python. Its flexible backend allows using NumPy, MXNet, PyTorch, TensorFlow or CuPy to execute the computations efficiently on various hardware platforms, and to use tensor methods in combination with deep learning.
With a prolific and impactful career in academia, in parallel to high-profile positions in industry, one wonders how it is possible for Anima to also be a science communicator and activist for both responsible AI and diversity and inclusion in tech. But she is all these things, and with utmost grace.
Inviting keynote speakers for JupyterCon is both a nerve-racking and gleeful duty for the organizers. Our aspiration is to experience together an unforgettable moment of intellectual joy and inspiration. And that, dear reader, is a guarantee with our first advertised speaker: Professor Anima Anandkumar.
— Lorena A Barba, JupyterCon 2020 General Chair
