LebNet
LebNet Bay Area - AI and the Future of Innovation Panel Discussion
An AI panel with Hadi Salman, Rima Arnaout, and Nader Khalil on industries, work, safety, and emerging capabilities.
Research in the world
Featured work & media.
Interviews, talks, videos, press coverage, research stories, awards, and community news.
65 entries · Newest first
American University of Beirut
AI Foundations — AUB North American Technology Conference
An AI Foundations panel at AUB’s North American Technology Conference in Silicon Valley.
ACM AUB
Careers in AI & Big Tech
A student panel about careers in AI research and technology with Hadi Kotaich, moderated by Ammar Mohanna.
LebNet Bay Area
6th Bireme Awards panel
Moderated a discussion with award winners Simon Kalouche and Alaa Khaddaj alongside Aya Mouallem.
LebNet
Hadi Salman Joins the LebNet Board
LebNet welcomes Hadi as an Associate Board Member, highlighting his AI research and community-building work.
LebNet Bay Area
Bireme Awards panel
A discussion of research, professional journeys, and the Lebanese technology community with the Bireme Award honorees.
KAUST Rising Stars in AI
Adversarial Examples Beyond Security
A research talk on uses of adversarial examples beyond security.
AUB · MSFEA IDEAS Awards
Emerging Trailblazer Award
Recognized by AUB’s Maroun Semaan Faculty of Engineering and Architecture in Mechanical Engineering.
Muslims in ML · NeurIPS
Adversarial Examples Beyond Security
A keynote on beneficial uses of adversarial examples, from robust objects to protection against unwanted image edits.
Asharq Al-Awsat
PhotoGuard: protecting images from AI manipulation
An Arabic interview on PhotoGuard, image manipulation, and adoption by AI developers.
MIT CSAIL
Towards ML Models That We Can Deploy Confidently
A thesis defense on adversarial robustness, reliable model deployment, debugging, and protection against unwanted AI image edits.
MIT CSAIL
Using AI to protect against AI image manipulation
A conversation about PhotoGuard and protecting images from unwanted AI manipulation.
Boston.com
MIT researchers create new tool to fight AI image manipulation
Coverage of PhotoGuard and the practical challenges of protecting images against AI manipulation.
MIT Technology Review Korea
날로 심각해지는 딥페이크 포르노로부터 나를 지켜주는 도구들
A broader article discussing PhotoGuard and Glaze as tools for protecting images.
CNN
‘It gave us some way to fight back’: New tools aim to protect art and images from AI’s grasp
A feature on tools protecting images and artwork, including PhotoGuard.
The Register
MIT bods offer PhotoGuard gadget to thwart AI deepfakes
A technical report explaining the two attack methods and PhotoGuard's practical limits.
MIT News
Using AI to protect against AI image manipulation
MIT's feature on PhotoGuard, its two protection methods, and the collaboration needed for deployment.
VentureBeat
MIT CSAIL unveils PhotoGuard, an AI defense against unauthorized image manipulation
Interview about PhotoGuard’s image-protection approach.
I Programmer
PhotoGuard - Protecting Images From Misuse By Generative AI
An explainer connecting PhotoGuard's defense to adversarial perturbations, with a demonstration.
Creative Bloq
This tool could save your photos from AI (for now)
A design-focused feature on PhotoGuard and its potential to protect personal photographs.
TecMundo
PhotoGuard: ferramenta criada por pesquisadores pode proteger as suas fotos de manipulação por IA
Brazilian Portuguese coverage of the research and its protection against AI photo manipulation.
Gadgets Now / Times of India
Explained: How this new tool can protect your images from AI edits
An explainer on PhotoGuard's image protection and practical limitations.
MIT Technology Review
This new tool could protect your pictures from AI manipulation
A feature on PhotoGuard, image protection, and the role of AI developers.
WIRED Italia
Lo strumento del MIT che impedisce all'AI di modificare le nostre foto
An Italian feature explaining PhotoGuard and Hadi's work on protection against malicious image edits.
Gizmodo en Español
Esta herramienta del MIT protege tus fotos de retoques maliciosos basados en IA
Spanish coverage of PhotoGuard's protection technique and the need for industry participation.
t3n
Deepfake-Sperre: Diese Software schützt eure Fotos vor Manipulation
German coverage explaining PhotoGuard's methods and the risks of manipulated photos.
DesignTAXI
MIT Creates ‘Photoguard’ Tool To Prevent Your Photos From AI Manipulation
Design coverage illustrating how protected photos resist AI edits.
Gradient Science
Rethinking Backdoor Attacks
A research-team explainer on assumptions behind backdoor detection and an approach based on data attribution.
PetaPixel
‘Photoguard’ Stops Your Pictures Being Manipulated by AI
An early photography-focused feature showing PhotoGuard's image-editing examples.
ICML 2023
Raising the Cost of Malicious AI-Powered Image Editing
A conference paper presentation on making images harder to manipulate with generative AI.
ICML 2023
Rethinking Backdoor Attacks
A conference paper presentation examining assumptions behind backdoor attacks and defenses.
CVPR 2023
FFCV: Accelerating Training by Removing Data Bottlenecks
A conference paper presentation on accelerating model training by reducing data bottlenecks.
CVPR 2023
A Data-Based Perspective on Transfer Learning
A conference paper presentation on understanding transfer learning through the training data.
Ars Technica
AI image generation tech can now create life-wrecking deepfakes with ease
A wider report on deepfake risks that discusses PhotoGuard as a possible defense against image editing.
Microsoft Research · NeurIPS
3DB: A Framework for Debugging Computer Vision Models
A live research demonstration with Hadi Salman and Sai Vemprala on debugging computer vision models.
GIGAZINE
画像に「目に見えない改変」を加えてAIによる画像加工を防ぐ「PhotoGuard」が開発される
An early Japanese report demonstrating how PhotoGuard disrupts AI image editing.
Gizmodo
Who Is Working to End the Threat of AI-Generated Deepfakes, and Why Is It So Difficult?
Interview on image protection.
ZAKA
Adversarial Attacks and Defenses in Machine Learning
A masterclass introducing adversarial examples, attacks, and defense strategies.
Microsoft Research
3DB: Debugging Computer Vision Models through Simulation
Using photorealistic simulation to discover and analyze failures caused by object pose, background, camera effects, and other scene changes.
Gradient Science
A Data-Based Perspective on Transfer Learning
Tracing downstream model behavior back to pretraining data to identify harmful examples, subpopulations, and data leakage.
Gradient Science
When does Bias Transfer in Transfer Learning?
Examining how biases in pretrained models can persist after adaptation to a carefully curated downstream dataset.
NeurIPS 2022
3DB: A Framework for Debugging Computer Vision Models
A conference paper presentation on using simulation to debug computer vision models.
ICLR 2022
Missingness Bias in Model Debugging
A conference paper presentation on the effects of missing inputs when debugging models.
SRML Workshop · ATVA
Certified Patch Robustness via Smoothed Vision Transformers
A research talk on certified protection against adversarial patches using smoothed vision transformers.
Gradient Science
Certified Patch Robustness Via Smoothed Vision Transformers (Part 1)
An introduction to randomized smoothing and guarantees against bounded adversarial image patches.
Gradient Science
Certified Patch Robustness Via Smoothed Vision Transformers (Part 2)
How vision transformers improve certified patch robustness and the speed of smoothed prediction.
Gradient Science
3DB: A Framework for Debugging Vision Models
The research team's walkthrough of a simulation framework for finding and understanding computer vision failures.
Machine Learning Street Talk
#52 — Adversarial Examples Beyond Security
A podcast conversation about adversarial robustness, robust objects, and transfer learning.
Microsoft Research
Denoised smoothing: Provably defending pretrained classifiers against adversarial examples
A method for adding robustness guarantees to existing image classifiers using a denoiser and randomized smoothing.
NeurIPS 2021
Unadversarial Examples: Designing Objects for Robust Vision
A conference paper presentation on designing objects that vision models recognize more reliably.
Microsoft Research
Unadversarial examples: Designing objects for robust vision
How specially designed visual patterns can help vision models recognize objects across changes in orientation, corruptions, and simulated weather.
Gradient Science
Unadversarial Examples: Designing Objects for Robust Vision
The research team's project walkthrough and demonstrations of visual patterns designed for reliable machine recognition.
Microsoft Research
Adversarial robustness as a prior for better transfer learning
An explanation of how robust pretraining can improve downstream image classification, object detection, and segmentation.
Synced
Microsoft & MIT Apply Adversarially Robust Models for Better Transfer Learning
Coverage of research showing that adversarially robust ImageNet models can improve transfer to downstream vision tasks.
Microsoft Research · TRAC Workshop
Do Adversarially Robust ImageNet Models Transfer Better?
A lightning talk on transfer learning with adversarially robust ImageNet models.
Decent Descent
Randomized Smoothing of All Shapes and Sizes, Part I
An explanation of how the choice of smoothing distribution can produce robustness guarantees for different kinds of perturbations.
NeurIPS 2020
Do Adversarially Robust ImageNet Models Transfer Better?
A conference paper presentation on transfer learning with adversarially robust ImageNet models.
NeurIPS 2020
Denoised Smoothing: A Provable Defense for Pretrained Classifiers
A conference paper presentation on certifiable defenses for pretrained image classifiers.
ICML 2020
Randomized Smoothing of All Shapes and Sizes
A conference paper presentation on randomized smoothing with different noise distributions.
Microsoft Research
Provable guarantees come to the rescue to break attack-defense cycle in adversarial machine learning
Combining adversarial training and randomized smoothing to strengthen provable robustness for image classifiers.
Microsoft Research
AI Residency: 7 months in and looking ahead
An interview about the AI Residency, robust machine learning, and large-scale experimentation.
Decent Descent
Provably Robust Deep Learning
A technical introduction to adversarial training of smoothed classifiers and the resulting robustness guarantees.
ICAPS 2017
Multi-Agent Ergodic Coverage with Obstacle Avoidance
A conference paper presentation on coordinating robot coverage while avoiding obstacles.
No entries match your search. Try another topic or category.