Machine Learning News | Nov 27, 2022

Machine Learning News | Nov 27, 2022

Featured Post?????????

This Invisible Sweater Developed by the University of Maryland Tricks?Artificial Intelligence (AI) Cameras?and Stops them from Recognizing People

No alt text provided for this image

Researchers show a systematic study of how adversarial attacks on the best object detection frameworks can be moved to other frameworks. They train patterns that suppress the objectness scores made by various commonly used detectors and groups. They do this by using standard detection datasets. Through much testing, researchers found how well adversarially trained patches work in both white-box and black-box settings and how well attacks can be transferred between datasets, object classes, and detector models. Lastly, they show a detailed study of physical world attacks using printed posters and clothes worn, and they use different metrics to measure how well these attacks work.

Scientists have made a real-life “invisibility cloak” that fools artificial intelligence (AI) cameras and stops them from recognizing people. Researchers have made a sweater that “breaks” AI systems that recognize people and makes a person “invisible” to AI cameras.

According to the research group, this stylish sweater is a great way to stay warm this winter. “It has a modern cut, a waterproof microfleece lining, and anti-AI patterns that will help you hide from object detectors.”

The researchers say, “In their demonstration, they were able to trick the YOLOv2 detector by using a pattern trained on a COCO data set and a carefully made target.”


Trending AI/ML Research ????????

No alt text provided for this image

???Meet TAP-Vid: A Dataset of Videos Along With Point Tracks, Either Manually Annotated or Obtained From A Simulator

???Google AmbiML Open-Sources ‘KataOS,’ A Secure Operating System For Embedded Machine Learning Hardware

???Meet ‘EDGE’: A Diffusion Based AI Model That Generates Realistic And Long-Form Dance Sequences Conditioned On Music

???Researchers At Stanford Have Developed A New Artificial Intelligence (AI) Benchmark To Understand Large Language Models (LLMs)

???Meet MultiRay: Meta AI’s New Platform For Efficiently Running Large-Scale Artificial Intelligence (AI) Models

???Amazon AI Researchers Propose A New Machine Learning Framework Called ‘GRAVL-BERT’: A BERT-Based Graphical Visual-Linguistic Representations For Multimodal Coreference Resolution


MLOps News ????????

???MakinaRocks Unveils “Runway” – Targeting the Enterprise MLOps Market

???MLOps platform Galileo lands $18M to launch a free service ?

???A New MLOps System Called ALaaS (Active-Learning-as-a-Service) Adopts the Philosophy of Machine-Learning-as-Service and Implements a Server-Client Architectur

???Top Tools To Log And Manage Machine Learning Models


Cool Papers ????????

???Paint by Example: Exemplar-based Image Editing with Diffusion Models

???TorchScale: Transformers at Scale

???DiffusionDet: Diffusion Model for Object Detection

???AltCLIP: Altering the Language Encoder in CLIP .....

???SinDiffusion: Learning a Diffusion Model from a Single Natural Image

???EVA: Exploring the Limits of Masked Visual Representation Learning at Scale

Khalil Adib

Data Scientist @ Firemind | Applied AI and Machine Learning Solutions | x6 AWS certified

1 年

要查看或添加评论,请登录

社区洞察

其他会员也浏览了