ABAB Pedia

Jeffrey Pennington

Jeffrey Pennington is a machine learning researcher whose work moved from theoretical particle physics to word representations and deep learning theory. He co-authored GloVe with Richard SocherRichard SocherRichard Socher is an AI researcher, entrepreneur and investor. He is co-founder and CEO of Recursive, founder and CEO of You.com, and co-founder and managing partner of AIX Ventures. A Stanford computer science PhD, he contributed to the Stanford Sentiment Treebank and GloVe, founded MetaMind, and served as chief scientist at Salesforce.Open the full entry and Christopher ManningChristopher ManningChristopher Manning is a natural language processing researcher and Stanford professor emeritus, a former Stanford AI Lab director and an AIX Ventures investor. He collaborated with Richard Socher on recursive neural networks, sentiment analysis and word vectors.Open the full entry.

Contents6 sections
Key facts

Physics and mathematics education

From 2013: Moving from particle physics to NLP

He conducted doctoral research at SLAC and subsequently joined the Stanford NLP group as a postdoctoral researcher studying word representations and deep learning algorithms.⁠[2][1]

GloVe word vectors

He co-authored GloVe with Richard Socher and Christopher Manning, learning word vectors from global co-occurrence statistics.⁠[3]

Nonlinear random matrix research

Pennington and Pratik Worah studied how nonlinear activations can be incorporated into random-matrix analysis, using spectral distributions to investigate random-feature models and neural networks.⁠[4]

High-dimensional statistics and deep learning

A 2021 University of Colorado Boulder lecture listed him as a Staff Research Scientist at Google BrainGoogle BrainGoogle Brain was Google’s AI research team, co-founded by Jeff Dean and others in 2011. It open-sourced the TensorFlow machine learning system in 2015 and was merged with DeepMind into Google DeepMind in April 2023.Open the full entry. His talk examined bias, variance and generalization in high-dimensional models, connecting random features and neural tangent kernels with neural-network training.⁠[2]

Sources

  1. Jeffrey Pennington — Curriculum Vitae (opens in a new window)
  2. APPM Department Colloquium - Jeffrey Pennington (opens in a new window)
  3. GloVe: Global Vectors for Word Representation (opens in a new window)
  4. Nonlinear random matrix theory for deep learning (opens in a new window)