Jacob Kahn

Research Engineer

Facebook AI Research

contact[at]jacobkahn[dot]me

About

My research interests are in machine learning and artificial intelligence. I'm broadly interested in building compute and data-efficient models and frameworks. My background is in math and computer systems.

EXPERIENCE

Current: I'm a Research Engineer at Facebook AI Research (FAIR), where I built Flashlight, a high-performance, standalone library for machine learning, which emerged from wav2letter, a research framework for automatic speech recognition.


You can find my Facebook Research page here, my Facebook AI page here, and my publications below.

EDUCATION

I graduated from the M&T Program at the University of Pennsylvania in 2018, where I was advised by and worked under Boon Thau Loo, CJ Taylor, and Dan Roth. While at Penn, I was one of the directors of PennApps. I graduated with the following degrees:

TEACHING

I've TA'ed many core undergraduate and graduate computer science courses at Penn.

(*) head teaching assistant | (^) graduate course.

  • CIS 320: Algorithms*

    (Spring 2017, Fall 2017, Spring 2018)

    Combinatorial optimization, sorting, hashing, graphs, complexity theory.

  • CIS 505: Distributed Systems^

    (Spring 2018)

    Synchronization, communication, replication. Project-based, in C++.

  • CIS 550: Database and Information Systems^

    (Spring 2017)

    Structured data, modeling, architecture, and distributed processing.

  • CIS 548: Operating Systems*^

    (Spring 2018)

    Concurrency, resource management, virtual memory, file systems, virtual machines.

  • CIS 240: Introduction to Computer Architecture

    (Fall 2015, Fall 2016)

    Hardware structures, organization, machine language, C.

  • CIS 700: Computer Systems^

    (Spring 2017)

    Experimental course in advanced topics in operating systems.

  • CIT 593: Introduction to Computer Systems^

    (Fall 2016)

    Digital logic and circuits, von Neumann architecture, ISAs, C.

  • CIS 380: Operating Systems

    (Fall 2017)

    Introductory operating system design and implementation. Project-based, in C.

Selected Publications

2022

Flashlight: Enabling Innovation in Tools for Machine Learning

J. Kahn

,

V. Pratap

,

T. Likhomanenko

,

Q. Xu

,

A. Hannun

,

J. Cai

,

P. Tomasello

,

A. Lee

,

E. Grave

,

G. Avidov

,

B. Steiner

,

V. Liptchinsky

,

G. Synnaeve

,

R. Collobert

International Conference on Machine Learning (ICML) 2022 (Spotlight Oral)

2021

Robust wav2vec 2.0: Analyzing Domain Shift in Self-Supervised Pre-Training

W. Hsu

,

A. Sriram

,

A. Baevski

,

T. Likhomanenko

,

Q. Xu

,

V. Pratap

,

J. Kahn

,

A. Lee

,

R. Collobert

,

G. Synnaeve

,

M. Auli

Interspeech 2021

slimIPL: Language-Model-Free Iterative Pseudo-Labeling

Tatiana Likhomanenko*

,

Qiantong Xu*

,

Jacob Kahn

,

Gabriel Synnaeve

,

Ronan Collobert

Interspeech 2021

(*) Equal contribution.

Rethinking Evaluation in ASR: Are Our Models Robust Enough?

Tatiana Likhomanenko*

,

Qiantong Xu*

,

Vineel Pratap

,

Paden Tomasello

,

Jacob Kahn

,

Gilad Avidov

,

Ronan Collobert

,

Gabriel Synnaeve

Interspeech 2021

(*) Equal contribution.

2020

Libri-Light: A Benchmark for ASR with Limited or No Supervision

J. Kahn*

,

M. Rivière*

,

W. Zheng*

,

E. Kharitonov*

,

Q. Xu*

,

P.E. Mazaré*

,

J. Karadayi*

,

V. Liptchinsky

,

R. Collobert

,

C. Fuegen

,

T. Likhomanenko

,

G. Synnaeve

,

A. Joulin

,

A. Mohamed

,

E. Dupoux

ICASSP 2020

(*) Equal contribution.

Scaling Up Online Speech Recognition Using ConvNets

V. Pratap

,

Q. Xu

,

J. Kahn

,

G. Avidov

,

T. Likhomanenko

,

A. Hannun

,

V. Liptchinksy

,

G. Synnaeve

,

R. Collobert

Interspeech 2020

Iterative Pseudo-Labeling for Speech Recognition

Qiantong Xu

,

Tatiana Likhomanenko

,

Jacob Kahn

,

Awni Hannun

,

Gabriel Synnaeve

,

Ronan Collobert

Interspeech 2020

Self-Training for End-to-End Speech Recognition

Jacob Kahn

,

Ann Lee

,

Awni Hannun

ICASSP 2020

Differentiable Weighted Finite-State Transducers

Awni Hannun

,

Vineel Pratap

,

Jacob Kahn

,

Wei-Ning Hsu

Preprint; arXiv:2010.01003

2019

End-to-End ASR: from Supervised to Semi-Supervised Learning with Modern Architectures

Gabriel Synnaeve*

,

Qiantong Xu*

,

Jacob Kahn*

,

Edouard Grave*

,

Tatiana Likhomanenko

,

Vineel Pratap

,

Anuroop Sriram

,

Vitaliy Liptchinsky

,

Ronan Collobert*

ICML Workshop on Self-Supervision in Audio and Speech

(*) Equal contribution.

2018

wav2Letter++: A Fast Open-source Speech Recognition System

Vineel Pratap

,

Awni Hannun

,

Qiantong Xu

,

Jeff Cai

,

Jacob Kahn

,

Gabriel Synnaeve

,

Vitaliy Liptchinsky

,

Ronan Collobert

ICASSP 2019

Projects

MultiplayAR: Multiplayer Augmented Reality

Thesis | Penn Engineering

PENN

RESEARCH

DEV

We develop new and leverage modern computer vision algorithms to build a Unity framework to enable ARKit developers to incorporate multiplayer views into their augmented reality applications. Selected as one of the top three theses in Computer Science. Advised by Prof. CJ Taylor.

Relative Performance of [Un]structured Classifiers for Synthetic Prediction

CIS 700: ML for NLP

PENN

RESEARCH

DEV

We consider a simple synthetic sequence-to-sequence learning task and compare the performance of structured-global and unstructured methods. We compare performance over output space complexity, the amount of training data available, and the effects of added noise.