Researcher

Avrim Blum

Also known as Avrim L. Blum

Avrim Blum is a computer scientist whose work spans theoretical computer science and machine learning. He earned degrees at MIT and served on Carnegie Mellon’s computer science faculty from 1992 to 2017, with research in learning theory, approximation algorithms, privacy, and fairness. With Tom Mitchell, he published the 1998 co-training paper that initiated multi-view learning; later, he developed a polynomial-time co-training algorithm with Yishay Mansour for linear separators. He became an ACM Fellow in 2007, received CMU’s Herbert A. Simon Award for Teaching Excellence in Computer Science, and as of October 2026 serves as TTIC’s interim president and chief academic officer.

Photo of Avrim Blum

Timeline

  1. 1987–1991

    Early results in machine learning

    At MIT, he earned bachelor’s, master’s and PhD degrees in 1987, 1989 and 1991. In 1988, he published “Training a 3-node neural network is NP-complete.”

  2. 1992–2017

    From theory to fairness

    He served on Carnegie Mellon’s computer science faculty from 1992 to 2017. His research spanned theoretical computer science and machine learning, including learning theory, approximation algorithms, algorithmic game theory, privacy and algorithmic fairness.

  3. 1998

    Learning from two data sources

    With Tom Mitchell, he published the 1998 paper “Combining Labeled and Unlabeled Data with Co-Training,” which initiated the area of multi-view learning.

  4. 2007

    An ACM fellowship

    The ACM named him a Fellow in 2007 for contributions to computational learning theory, approximation algorithms, online algorithms and AI planning.

  5. 2011

    Teaching computer science

    He received CMU’s Herbert A. Simon Award for Teaching Excellence in Computer Science.

  6. 2017

    Co-training under weak dependence

    With Yishay Mansour, he coauthored the first polynomial-time algorithm for co-training homogeneous linear separators under weak dependence. It learns from unlabeled data with one labeled example to break class symmetry.

  7. As of October 2026

    Leading TTIC

    He joined TTIC as chief academic officer in 2017. At the institute’s September 2026 graduation ceremony, he served as interim president and chief academic officer.

Sources and credits

Photo: Rweba, via Wikimedia Commons. License: CC BY-SA 3.0. Source: commons.wikimedia.org. Cropped to a square and resized. Ask us to remove or replace this photo.

  1. TTIC Dr. Avrim Blum ttic.edu. Checked 14 October 2026.
  2. TTIC News ttic.edu. Checked 14 October 2026.
  3. Combining Labeled and Unlabeled Data with Co-Training | ML Anthology mlanthology.org. Checked 14 October 2026.
  4. SCS Faculty: Fellows of the Association for Computing Machinery (ACM) scs.cmu.edu. Checked 14 October 2026.
  5. Blum Awarded the Herbert A. Simon Award for Teaching Excellence ml.cmu.edu. Checked 14 October 2026.
  6. Efficient Co-Training of Linear Separators under Weak Dependence proceedings.mlr.press. Checked 14 October 2026.