Isaac Newton Institute Seminars

24
Apr

Algorithmic stability for heavy-tailed SGD

Speaker:

Lingjiong Zhu (Florida State University)

Time:

Wednesday, April 24, 2024 - 14:30 to 15:15

Venue:

External

24
Apr

Aggregating regularly varying vectors: phenomena of a few large jumps

Speaker:

Bikramjit Das (Singapore University of Technology and Design)

Time:

Wednesday, April 24, 2024 - 15:15 to 16:00

Venue:

External

25
Apr

Sharp Characterization and Control of Global Dynamics of SGDs with Heavy Tails

Speaker:

Xingyu Wang (Northwestern University)

Time:

Thursday, April 25, 2024 - 09:45 to 10:30

Venue:

External

25
Apr

Coffee Break

Speaker:

Time:

Thursday, April 25, 2024 - 10:30 to 11:00

Venue:

External

25
Apr

Stochastic Gradient Descent with Adaptive Data

Speaker:

Jing Dong (Columbia University)

Time:

Thursday, April 25, 2024 - 11:00 to 11:45

Venue:

External

25
Apr

Stochastic approximation with heavy tailed noise

Speaker:

Vivek Borkar (Indian Institute of Technology Bombay)

Time:

Thursday, April 25, 2024 - 11:45 to 12:30

Venue:

External

25
Apr

Lunch

Speaker:

Time:

Thursday, April 25, 2024 - 12:30 to 14:30

Venue:

External

25
Apr

Quantitative and stable limits of high-frequency statistics of L\'evy processes: a Stein's method approach

Speaker:

Chiara Amorino (Université du Luxembourg)

Time:

Thursday, April 25, 2024 - 14:30 to 15:15

Venue:

External

25
Apr

Emergence of heavy tails in homogenised stochastic gradient descent

Speaker:

Martin Keller-ressel (Technische Universität Dresden)

Time:

Thursday, April 25, 2024 - 15:15 to 16:00

Venue:

External

26
Apr

Welcome and Introduction

Speaker:

Jorge Gonzalez-Cazares (National Autonomous University of Mexico (UNAM), IIMAS), Sofia Sanz Del Pino (Isaac Newton Institute)

Time:

Friday, April 26, 2024 - 10:30 to 10:40

Venue:

External

26
Apr

Implicit Compressibility of Overparametrized Neural Networks Trained with Heavy-Tailed SGD

Speaker:

Umut Simsekli (INRIA)

Time:

Friday, April 26, 2024 - 10:40 to 11:20

Venue:

External

26
Apr

EMG reading group on Gopakuma-Vafa invariants.

Speaker:

Time:

Friday, April 26, 2024 - 11:00 to 13:00

Venue:

Seminar Room 1, Newton Institute

26
Apr

Refreshment break

Speaker:

Time:

Friday, April 26, 2024 - 11:20 to 11:35

Venue:

External

26
Apr

Generative Modelling Under Differential Privacy

Speaker:

Sahra Ghalebikesabi (Google)

Time:

Friday, April 26, 2024 - 11:35 to 12:15

Venue:

External

26
Apr

Lunch

Speaker:

Time:

Friday, April 26, 2024 - 12:15 to 13:00

Venue:

External

26
Apr

Moderated discussion: Heavy tails and clipping: benefits and difficulties.

Speaker:

Hao Ni (University College London)

Time:

Friday, April 26, 2024 - 13:00 to 13:35

Venue:

External

26
Apr

Enhancing Collaborative Medical Outcomes through Private Synthetic Hypercube Augmentation: PriSHA

Speaker:

Shinpei Nakamura (Yale University)

Time:

Friday, April 26, 2024 - 13:35 to 14:15

Venue:

External

26
Apr

title and abstract tba

Speaker:

Barbara Fantechi (SISSA)

Time:

Friday, April 26, 2024 - 14:00 to 15:00

Venue:

Seminar Room 1, Newton Institute

26
Apr

Refreshment break

Speaker:

Time:

Friday, April 26, 2024 - 14:15 to 14:30

Venue:

External

26
Apr

Privacy Properties of Discrete Diffusion Models

Speaker:

Vamsi Potluru (JP Morgan)

Time:

Friday, April 26, 2024 - 14:30 to 15:10

Venue:

External

24
Apr

Algorithmic stability for heavy-tailed SGD

24
Apr

Aggregating regularly varying vectors: phenomena of a few large jumps

25
Apr

Sharp Characterization and Control of Global Dynamics of SGDs with Heavy Tails

25
Apr

Coffee Break

25
Apr

Stochastic Gradient Descent with Adaptive Data

25
Apr

Stochastic approximation with heavy tailed noise

25
Apr

Lunch

25
Apr

Quantitative and stable limits of high-frequency statistics of L\'evy processes: a Stein's method approach

25
Apr

Emergence of heavy tails in homogenised stochastic gradient descent

26
Apr

Welcome and Introduction

26
Apr

Implicit Compressibility of Overparametrized Neural Networks Trained with Heavy-Tailed SGD

26
Apr

EMG reading group on Gopakuma-Vafa invariants.

26
Apr

Refreshment break

26
Apr

Generative Modelling Under Differential Privacy

26
Apr

Lunch

26
Apr

Moderated discussion: Heavy tails and clipping: benefits and difficulties.

26
Apr

Enhancing Collaborative Medical Outcomes through Private Synthetic Hypercube Augmentation: PriSHA

26
Apr

title and abstract tba

26
Apr

Refreshment break

26
Apr

Privacy Properties of Discrete Diffusion Models

Forthcoming Seminars

News, Announcements and Events

Study at Cambridge

About the University

Research at Cambridge

Isaac Newton Institute Seminars

24Apr Algorithmic stability for heavy-tailed SGD

24Apr Aggregating regularly varying vectors: phenomena of a few large jumps

25Apr Sharp Characterization and Control of Global Dynamics of SGDs with Heavy Tails

25Apr Coffee Break

25Apr Stochastic Gradient Descent with Adaptive Data

25Apr Stochastic approximation with heavy tailed noise

25Apr Lunch

25Apr Quantitative and stable limits of high-frequency statistics of L\'evy processes: a Stein's method approach

25Apr Emergence of heavy tails in homogenised stochastic gradient descent

26Apr Welcome and Introduction

26Apr Implicit Compressibility of Overparametrized Neural Networks Trained with Heavy-Tailed SGD

26Apr EMG reading group on Gopakuma-Vafa invariants.

26Apr Refreshment break

26Apr Generative Modelling Under Differential Privacy

26Apr Lunch

26Apr Moderated discussion: Heavy tails and clipping: benefits and difficulties.

26Apr Enhancing Collaborative Medical Outcomes through Private Synthetic Hypercube Augmentation: PriSHA

26Apr title and abstract tba

26Apr Refreshment break

26Apr Privacy Properties of Discrete Diffusion Models

Forthcoming Seminars

News, Announcements and Events

Study at Cambridge

About the University

Research at Cambridge

24
Apr

Algorithmic stability for heavy-tailed SGD

24
Apr

Aggregating regularly varying vectors: phenomena of a few large jumps

25
Apr

Sharp Characterization and Control of Global Dynamics of SGDs with Heavy Tails

25
Apr

Coffee Break

25
Apr

Stochastic Gradient Descent with Adaptive Data

25
Apr

Stochastic approximation with heavy tailed noise

25
Apr

Lunch

25
Apr

Quantitative and stable limits of high-frequency statistics of L\'evy processes: a Stein's method approach

25
Apr

Emergence of heavy tails in homogenised stochastic gradient descent

26
Apr

Welcome and Introduction

26
Apr

Implicit Compressibility of Overparametrized Neural Networks Trained with Heavy-Tailed SGD

26
Apr

EMG reading group on Gopakuma-Vafa invariants.

26
Apr

Refreshment break

26
Apr

Generative Modelling Under Differential Privacy

26
Apr

Lunch

26
Apr

Moderated discussion: Heavy tails and clipping: benefits and difficulties.

26
Apr

Enhancing Collaborative Medical Outcomes through Private Synthetic Hypercube Augmentation: PriSHA

26
Apr

title and abstract tba

26
Apr

Refreshment break

26
Apr

Privacy Properties of Discrete Diffusion Models