adversarially robust generalization requires more data
play

Adversarially Robust Generalization Requires More Data Ludwig - PowerPoint PPT Presentation

Dec 06, 2023 •2.36k likes •2.52k views

Adversarially Robust Generalization Requires More Data Ludwig Schmidt Shibani Santurkar Dimitris Tsipras Poster #31 Kunal Talwar Aleksander M dry Adversarial Examples [Szegedy, Zaremba, Sutskever, Bruna, Erhan, Goodfellow,

  1. Adversarially Robust Generalization Requires More Data Ludwig Schmidt Shibani Santurkar Dimitris Tsipras Poster #31 Kunal Talwar Aleksander M ą dry

  2. Adversarial Examples [Szegedy, Zaremba, Sutskever, Bruna, Erhan, Goodfellow, Fergus, 2013] [Biggio, Corona, Maiorca, Nelson, Srndic, Laskov, Giacinto, Roli, 2013]

  3. Adversarial Examples [Szegedy, Zaremba, Sutskever, Bruna, Erhan, Goodfellow, Fergus, 2013] [Biggio, Corona, Maiorca, Nelson, Srndic, Laskov, Giacinto, Roli, 2013] What makes adversarial examples a hard problem? This paper: perspective on sample complexity

  4. <latexit sha1_base64="TCezhzHnvwDoFMFoQ4EK/L9JVgU=">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</latexit> <latexit sha1_base64="TCezhzHnvwDoFMFoQ4EK/L9JVgU=">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</latexit> <latexit sha1_base64="TCezhzHnvwDoFMFoQ4EK/L9JVgU=">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</latexit> <latexit sha1_base64="TCezhzHnvwDoFMFoQ4EK/L9JVgU=">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</latexit> Standard vs Robust Generalization x,y ∼ D [ loss ( f ( x ) , y ) ] “Standard” Generalization: E

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

A Spectral View of Adversarially Robust Features Shivam Garg Vatsal Sharan * Brian Zhang *

A Spectral View of Adversarially Robust Features Shivam Garg Vatsal Sharan * Brian Zhang *

A Spectral View of Adversarially Robust Features Shivam Garg Vatsal Sharan * Brian Zhang * Gregory Valiant Stanford University What are adversarial examples? Adding small amount of well-crafted noise to the test data fools the classifier More

574 views • 14 slides
Adversarially Robust Optimization with Gaussian Processes Ilija Bogunovic, Jonathan Scarlett,

Adversarially Robust Optimization with Gaussian Processes Ilija Bogunovic, Jonathan Scarlett,

Adversarially Robust Optimization with Gaussian Processes Ilija Bogunovic, Jonathan Scarlett, Stefanie Jegelka, Volkan Cevher Conference on Neural Information Processing Systems (Dec 2018) Gaussian Process Optimization Optimum Non-robust

393 views • 12 slides
Better generalization with less data using robust gradient descent Matthew J. Holland 1 Kazushi

Better generalization with less data using robust gradient descent Matthew J. Holland 1 Kazushi

Better generalization with less data using robust gradient descent Matthew J. Holland 1 Kazushi Ikeda 2 1 Osaka University 2 Nara Institute of Science and Technology Distribution robustness In practice, the learner does not know what kind of data

893 views • 16 slides
Calibrated Surrogate Losses for Adversarially Robust Classification 1 The University of Tokyo

Calibrated Surrogate Losses for Adversarially Robust Classification 1 The University of Tokyo

Calibrated Surrogate Losses for Adversarially Robust Classification 1 The University of Tokyo 2 RIKEN AIP 3 University of Michigan Jul. 9 th - 12 th @ COLT 2020 Han Bao 1,2 Clayton Scott 3 Masashi Sugiyama 2,1 Adversarial Attacks 2

514 views • 16 slides
Optimal Statistical Guarantees for Adversarially Robust Gaussian Classification Chen Dan, Yuting

Optimal Statistical Guarantees for Adversarially Robust Gaussian Classification Chen Dan, Yuting

Optimal Statistical Guarantees for Adversarially Robust Gaussian Classification Chen Dan, Yuting Wei, Pradeep Ravikumar ICML 2020 Computer Science Department, Statistics Department, Machine Learning Department Carnegie Mellon University

403 views • 16 slides
Multi-target Voice Conversion without Parallel Data by Adversarially Learning Disentangled Audio

Multi-target Voice Conversion without Parallel Data by Adversarially Learning Disentangled Audio

Setting - 2 Task Setting - 1 Multi-target Voice Conversion without Parallel Data by Adversarially Learning Disentangled Audio Representations Method Ju-chieh Chou , Cheng-chieh Yeh, Hung-yi Lee, Lin-shan Lee Best student paper award nominated

532 views • 25 slides
Data Anonymization - Generalization Algorithms Li Xiong, Slawek Goryczka CS573 Data Privacy and

Data Anonymization - Generalization Algorithms Li Xiong, Slawek Goryczka CS573 Data Privacy and

Data Anonymization - Generalization Algorithms Li Xiong, Slawek Goryczka CS573 Data Privacy and Anonymity Generalization and Suppression Generalization Suppression Replace the value with a less Do not release a Z2 = {410**}

1.27k views • 52 slides
Making Generalization Robust Katrina Ligett HUJI &amp; Caltech joint with Rachel Cummings, Kobbi

Making Generalization Robust Katrina Ligett HUJI &amp; Caltech joint with Rachel Cummings, Kobbi

Making Generalization Robust Katrina Ligett HUJI &amp; Caltech joint with Rachel Cummings, Kobbi Nissim, Aaron Roth, and Steven Wu A model for science A model for science Hypothesis Learning Alg domain: contains all possible

904 views • 56 slides
Data Anonymization - Generalization Algorithms Li Xiong CS573 Data Privacy and Anonymity

Data Anonymization - Generalization Algorithms Li Xiong CS573 Data Privacy and Anonymity

Data Anonymization - Generalization Algorithms Li Xiong CS573 Data Privacy and Anonymity Generalization and Suppression Generalization Suppression Replace the value with a less Do not release a Z2 = {410**} value at all

790 views • 45 slides
Learning From Data Lecture 7 Approximation Versus Generalization The VC Dimension Approximation

Learning From Data Lecture 7 Approximation Versus Generalization The VC Dimension Approximation

Learning From Data Lecture 7 Approximation Versus Generalization The VC Dimension Approximation Versus Generalization Bias and Variance The Learning Curve M. Magdon-Ismail CSCI 4100/6100 recap: The Vapnik-Chervonenkis Bound (VC Bound) P [ |

353 views • 22 slides
The Common Core World A Generalization about Where Were Headed Phyllis Hinton, Adult Education

The Common Core World A Generalization about Where Were Headed Phyllis Hinton, Adult Education

The Common Core World A Generalization about Where Were Headed Phyllis Hinton, Adult Education and Literacy Consultant July 2013 PD Standard 8: Requires leadership at the state and local levels that promotes effective professional

600 views • 32 slides
Learning From Data Lecture 22 Neural Networks and Overfitting Approximation vs. Generalization

Learning From Data Lecture 22 Neural Networks and Overfitting Approximation vs. Generalization

Learning From Data Lecture 22 Neural Networks and Overfitting Approximation vs. Generalization Regularization and Early Stopping Minimizing E in More Efficienty M. Magdon-Ismail CSCI 4100/6100 recap: Neural Networks and Fitting the Data

464 views • 15 slides
HALI: Hierarchical Adversarially Learned Inference Negar Rostamzadeh ACM Webinar January 18,

HALI: Hierarchical Adversarially Learned Inference Negar Rostamzadeh ACM Webinar January 18,

HALI: Hierarchical Adversarially Learned Inference Negar Rostamzadeh ACM Webinar January 18, 2018 Negar Rostamzadeh HALI January 18, 2018 1 / 43 Hierarchical Adversarially Learned Inference Mohamed Ishmael Belghazi, Sai Rajeswar, Olivier

468 views • 43 slides
Adversarially Learned Representations for Information Obfuscation and Inference Martin Bertran 1

Adversarially Learned Representations for Information Obfuscation and Inference Martin Bertran 1

Adversarially Learned Representations for Information Obfuscation and Inference Martin Bertran 1 , Natalia Martinez 1 , Afroditi Papadaki 2 Qiang Qiu 1 , Miguel Rodrigues 2 , Galen Reeves 1 , Guillermo Sapiro 1 1. Duke University 2. University

1.03k views • 41 slides
On the Generalization of Subspace Detection in Unordered Multidimensional Data Leandro Augusto

On the Generalization of Subspace Detection in Unordered Multidimensional Data Leandro Augusto

On the Generalization of Subspace Detection in Unordered Multidimensional Data Leandro Augusto Frata Fernandes laffernandes@inf.ufrgs.br Manuel Menezes de Oliveira Neto oliveira@inf.ufrgs.br Advisor CG UFRGS Data Alignments 2 Data

1.1k views • 43 slides
Chapter 2 SPATIAL DATA INPUT SPATIAL DATA INPUT Analysis and modeling in a GIS requires input of

Chapter 2 SPATIAL DATA INPUT SPATIAL DATA INPUT Analysis and modeling in a GIS requires input of

Chapter 2 SPATIAL DATA INPUT SPATIAL DATA INPUT Analysis and modeling in a GIS requires input of relevant data. The data consist of two types: spatial data representing geographic features (points, lines and areas) and attribute data (descriptive

550 views • 26 slides
Rotating neutron stars with nonbarotropic thermal profile Giovanni Camelio with Tim Dietrich,

Rotating neutron stars with nonbarotropic thermal profile Giovanni Camelio with Tim Dietrich,

Rotating neutron stars with nonbarotropic thermal profile Giovanni Camelio with Tim Dietrich, Miguel Marques, and Stephan Rosswog PRD 100:123001 (2019) 2020 February 10 Compact objects for all @ Lund Motivation dynamical simulation

328 views • 9 slides
VHDL Simulation Using Mentor Graphics Modelsim SE VHDL Simulation Tools Mentor Graphics

VHDL Simulation Using Mentor Graphics Modelsim SE VHDL Simulation Tools Mentor Graphics

VHDL Simulation Using Mentor Graphics Modelsim SE VHDL Simulation Tools Mentor Graphics Modelsim PE Student Edition: free download for academic course work: http://model.com/content/modelsim-pe-student-edition-hdl-simulation Full

520 views • 22 slides
THE POWER OF THE TERMINATING CHASE Markus Krtzsch Maximilian Marx Sebastian Rudolph TU

THE POWER OF THE TERMINATING CHASE Markus Krtzsch Maximilian Marx Sebastian Rudolph TU

THE POWER OF THE TERMINATING CHASE Markus Krtzsch Maximilian Marx Sebastian Rudolph TU Dresden Download Paper Lisbon, ICDT 2019 Invited Tutorial Fig. 1: The Chase Part 1: Tuple-Generating Dependencies Part 1: Existential Rules

1.75k views • 146 slides
How T To M Make Y Your Course I Interactive and E Engaging Stephen Murgatroyd, PhD Chief

How T To M Make Y Your Course I Interactive and E Engaging Stephen Murgatroyd, PhD Chief

How T To M Make Y Your Course I Interactive and E Engaging Stephen Murgatroyd, PhD Chief Innovation Officer Contact North | Contact Nord www.contactnord.ca Who ho Am Am I I Teaching at the university level since 1973

446 views • 33 slides
A Walk Up the Stack Jean Bolot 1. Intersections with Franois First paper read First paper

A Walk Up the Stack Jean Bolot 1. Intersections with Franois First paper read First paper

A Walk Up the Stack Jean Bolot 1. Intersections with Franois First paper read First paper co-authored Last paper co-authored Most used paper I-thought-would-be-most-used paper Move Up the Stack From networks To usage Move Up the

569 views • 32 slides
Media Streaming: Pushing the Envelope Pl Halvorsen Today: Everywhere, to any device, anytime

Media Streaming: Pushing the Envelope Pl Halvorsen Today: Everywhere, to any device, anytime

Media Streaming: Pushing the Envelope Pl Halvorsen Today: Everywhere, to any device, anytime Segmented adaptive HTTP streaming works great! No, it is not perfect protocol optimizations for ON-OFF traffic fairness

303 views • 8 slides
Aims and objectives Our aims: Share the latest information and guidance on the temporary

Aims and objectives Our aims: Share the latest information and guidance on the temporary

Aims and objectives Our aims: Share the latest information and guidance on the temporary disapplication and modification of parts of the early years foundation stage (EYFS). Answer stakeholder questions as best we can. By the end of the

467 views • 21 slides
The 2009 Aqua Senior Review and Other Aqua Updates Claire L. Parkinson Aqua Project ScienBst

The 2009 Aqua Senior Review and Other Aqua Updates Claire L. Parkinson Aqua Project ScienBst

The 2009 Aqua Senior Review and Other Aqua Updates Claire L. Parkinson Aqua Project ScienBst Presented at the AIRS Science Team meeBng, Greenbelt, October 13, 2009 2009 Senior Review Purpose: Determine which of the reviewed missions warrant

448 views • 23 slides

More recommend