Rosta Windows
7 by brudgers | 1 comments on Hacker News.
Thursday, March 31, 2022
In Paris, a Wine Bar Inspired by Tokyo’s Jazz Cafes
By Unknown Author from NYT T Magazine https://ift.tt/IibCZdO
Faced with foreign pressure, Russians rally around Putin, poll shows.
By BY IVAN NECHEPURENKO from NYT World https://ift.tt/7CqrYH2
Wednesday, March 30, 2022
‘Told It Like It Was’: Ntozake Shange’s Tales of Black Womanhood
By BY SOYICA DIGGS COLBERT from NYT Theater https://ift.tt/IDZAjo3
Hollywood Star Gives Broadway a Much-Needed Boost. Sound Familiar?
By BY LAURENCE MASLON from NYT Theater https://ift.tt/dP4beFA
The Nuclear Family Is No Longer the Norm. Good.
By BY JESSICA GROSE from NYT Opinion https://ift.tt/dgeqOWA
California Task Force Votes to Offer Reparations Only to Descendants of Enslaved People
By BY SOUMYA KARLAMANGLA from NYT U.S. https://ift.tt/DxbRiWc
Two U.K. Judges Quit Hong Kong Court, Citing Lost Freedoms
By BY AUSTIN RAMZY from NYT World https://ift.tt/m0ybsRC
Tuesday, March 29, 2022
Where Breaking the Ramadan Fast Includes Caribou
By BY VICTORIA PETERSEN from NYT Food https://ift.tt/nl0ELR5
How Shanghai’s lockdown is testing China’s ‘Covid-zero’ policy, and people’s limits.
By BY JOHN LIU AND PAUL MOZUR from NYT World https://ift.tt/0ium6XA
Daddy Yankee, Reggaeton’s First Global Star, Steps Aside
By BY ISABELIA HERRERA from NYT Arts https://ift.tt/Uat0qdR
More Private Jets Take to the Skies, Creating Gridlock on the Ground
By BY DEBRA KAMIN from NYT Business https://ift.tt/72WEouy
Experts warn of racial disparities in the diagnosis and treatment of long Covid.
By BY LOLA FADULU from NYT Health https://ift.tt/uVk0Qw4
California Hopes to Balance Legislation From Conservative States
By BY SOUMYA KARLAMANGLA from NYT U.S. https://ift.tt/IKRBkjG
Monday, March 28, 2022
South Korea reports a decline in the average in new cases and says the worst of Omicron may be over.
By BY JIN YU YOUNG from NYT World https://ift.tt/GPQSndH
Blinken says peace efforts will break down barriers between leaders, citizens and societies.
By BY LARA JAKES from NYT World https://ift.tt/6eM2P3n
Sunday, March 27, 2022
To ward off pandemic boredom, people across the U.S. walked every street in their city.
By BY MITCH SMITH from NYT U.S. https://ift.tt/yrEmbwo
N.C.A.A. Women’s Tournament: What to Watch in the Elite Eight
By BY NATALIE WEINER from NYT Sports https://ift.tt/SPGXKQE
Amid War, a Lavish Hotel Will Open in Britain’s Old War Office
By BY MARK LANDLER from NYT World https://ift.tt/QytLH4c
New top story on Hacker News: Mach v0.1 – cross-platform Zig graphics in ~60 seconds
Mach v0.1 – cross-platform Zig graphics in ~60 seconds
7 by andreafeletto | 1 comments on Hacker News.
7 by andreafeletto | 1 comments on Hacker News.
Ukraine says Russia struck a Holocaust memorial.
By BY EMMA BUBOLA from NYT World https://ift.tt/5pQEfW0
Saturday, March 26, 2022
Descendants Trace Histories Linked by Slavery
By BY AMANDA HOLPUCH from NYT U.S. https://ift.tt/0Z1qlzC
Getting Covid Shots in the Littlest Arms Might Not Be That Easy
By BY JESSICA GROSE from NYT Opinion https://ift.tt/4o3ks7f
Biden stresses the need for ‘a united Europe’ in a meeting with Poland’s president.
By BY MICHAEL D. SHEAR from NYT World https://ift.tt/GCSv9wN
Arizona Offers Driver’s Licenses on iPhones. Other States Want to Be Next.
By BY NEIL VIGDOR from NYT U.S. https://ift.tt/abRpOCK
Russian forces entered a city near Chernobyl and detained the mayor, officials say.
By BY MEGAN SPECIA from NYT World https://ift.tt/KLf6xXU
Addressing Qatar, Zelensky urges energy producers to increase their exports to Europe.
By BY MATTHEW MPOKE BIGG from NYT World https://ift.tt/V3TIRDi
Friday, March 25, 2022
Why Californians Have Been Saving Less Water in 2022
By BY SOUMYA KARLAMANGLA from NYT U.S. https://ift.tt/qE30VYa
Amazon is trying to fend off unions at two of its warehouses.
By Unknown Author from NYT Business https://ift.tt/YlLUnzP
Thursday, March 24, 2022
Homes for Sale in Brooklyn and Manhattan
By BY C. J. HUGHES from NYT Real Estate https://ift.tt/L869Bn0
Inside the Maximalist Homes of a Countess
By Unknown Author from NYT T Magazine https://ift.tt/Qw2UHZP
How Do I Tell My Neighbor to Stop Tending to My Lawn?
By BY PHILIP GALANES from NYT Style https://ift.tt/WDzKEkI
Ukrainians pouring into tiny Moldova find a welcoming but strained country.
By BY EMILY RHYNE, SIMON OSTROVSKY AND BEN LAFFIN from NYT World https://ift.tt/ua35V79
Wednesday, March 23, 2022
‘Mariupol is no more’: A Greek diplomat laments the city’s destruction after evacuating.
By BY NIKI KITSANTONIS from NYT World https://ift.tt/2AHwsgK
E.U. weighs sweeping energy market measures to tackle price and supply blows.
By BY MATINA STEVIS-GRIDNEFF from NYT World https://ift.tt/8USfZ0L
New top story on Hacker News: Show HN: We made an open-source personalization engine
Show HN: We made an open-source personalization engine
10 by shutty | 0 comments on Hacker News.
Hey, HN! You probably know that the ordering of products on Amazon, posts in FB, and search results in Google is personalized for each visitor, as it directly affects conversion, click rate and engagement. But not everyone can afford to hire an army of PhDs to squeeze every penny out of the ranking, and not everyone agrees on the current (im)balance between privacy and profits. So we built Metarank, an open-source and privacy-focused personalization engine. It can rerank in real-time any type of content, using only the data you allow, and optimize metrics you define. We made a lot of proprietary DIY services for personalization in e-commerce in our past careers and heard so many complaints from other companies also struggling to implement personalization. It’s often considered "too risky" to spend 6+ months on an in-house moonshot project to reinvent the wheel without an experienced team and no existing open-source tools. Like other people in the industry, we were tired of building everything from the bottom up each time we approached personalization - it should be easy not only for Amazon to do such magical ML tricks, but for everyone else. A small demo of the tool with personalized recommendations: https://ift.tt/5Tth8EZ A blog post on how this demo was made: https://ift.tt/aR0LPYU... The project itself: https://ift.tt/1RNcXZw
10 by shutty | 0 comments on Hacker News.
Hey, HN! You probably know that the ordering of products on Amazon, posts in FB, and search results in Google is personalized for each visitor, as it directly affects conversion, click rate and engagement. But not everyone can afford to hire an army of PhDs to squeeze every penny out of the ranking, and not everyone agrees on the current (im)balance between privacy and profits. So we built Metarank, an open-source and privacy-focused personalization engine. It can rerank in real-time any type of content, using only the data you allow, and optimize metrics you define. We made a lot of proprietary DIY services for personalization in e-commerce in our past careers and heard so many complaints from other companies also struggling to implement personalization. It’s often considered "too risky" to spend 6+ months on an in-house moonshot project to reinvent the wheel without an experienced team and no existing open-source tools. Like other people in the industry, we were tired of building everything from the bottom up each time we approached personalization - it should be easy not only for Amazon to do such magical ML tricks, but for everyone else. A small demo of the tool with personalized recommendations: https://ift.tt/5Tth8EZ A blog post on how this demo was made: https://ift.tt/aR0LPYU... The project itself: https://ift.tt/1RNcXZw
N.Y.C.’s New Subway Chief Comes From Boston and Doesn’t Own a Car
By BY MICHAEL GOLD from NYT New York https://ift.tt/QTPg4kM
Tuesday, March 22, 2022
On Day 2 of hearing, two sides will draw different portraits of Jackson.
By BY CHARLIE SAVAGE from NYT U.S. https://ift.tt/nLOTHgB
New top story on Hacker News: Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform
Launch HN: Reality Defender (YC W22) – Deepfake Detection Platform
1 by bpcrd | 0 comments on Hacker News.
Hi HN, we’re Ben, Gaurav and Ali from Reality Defender ( https://ift.tt/4krQmD8 ). We help companies, governments, and journalists determine if media is real or fake, focusing on audio, video and image manipulation. Our API and web app provide real-time scanning, risk scoring, and PDF report cards. Recent advancements in machine learning make it possible to create images, videos and audio of real people saying and doing things they never said or did. The recent spread of this technology has enabled anyone to create highly realistic deepfakes. Although some deepfakes are detectable to the eye by experienced observers who look closely, many people either don’t have experience or are not always looking closely—and of course the technology is only continuing to improve. This marks a leap in the ability of bad actors to distort reality, jeopardizing financial transactions, personal and brand reputations, public opinion, and even national security. We are a team with PhD and Master degrees from Harvard, NYU and UCLA in data science. Between us, we have decades of experience at Goldman Sachs, Google, CIA, FDIC, Dept of Defense and Harvard University Applied Research at the intersection of machine learning and cybersecurity. But our current work began with a rather unlikely project: we tried to duplicate Deepak Chopra. We were working with him to build a realistic deepfake that would allow users to have a real-time conversation with “Digital Deepak” from their iPhones. Creating the Deepak deepfake was surprisingly simple and the result was so alarmingly realistic that we immediately began looking for models that could help users tell a synthetic version from the real thing. We did not find a reliable solution. Frustrated that we’d already spent a week on something we thought would take our coffee break, we doubled down and set out to build our own model that could detect manipulated media. After investigating, we learned why a consistently accurate solution didn’t exist. Companies (including Facebook and Microsoft) were trying to build their own silver-bullet, single-model detection methods—or, as we call it, "one model to rule them all." In our view, this approach will not work because adversaries and the underlying technologies are constantly evolving. For this same reason there will never be a single model to solve anti-virus, malware, etc. We believe that any serious solution to this problem requires a “multi-model'' approach that integrates the best deepfake detection algorithms into an aggregate "model of models." So we trained an ensemble of deep-learning detection models, each of which focuses on its own feature, and then combined the scores. We challenged ourselves to build a scalable solution that integrates the best of our deepfake detection models with models from our collaborators (Microsoft, UC Berkeley, Harvard). We began with a web app proof of concept, and quickly received hundreds of requests for access from governments, companies, and researchers. Our first users turned to our platform for some deepfake scenarios ranging from bad to outright scary: Russian disinformation directed at Ukraine and the West; audio mimicking a bank executive requesting a wire transfer; video of Malaysia’s government leadership behaving scandalously; pornography where participants make themselves appear younger; dating profiles with AI-generated pro pics. All of these, needless to say, are completely fake! As with computer viruses, deepfakes will continue evolving to circumvent current security measures. New deepfake detection techniques must be as iterative as the generation methods. Our solution not only accepts that, but embraces it. We quickly onboard, test, and tune third party models for integration into our model stack, where they can then be accessed via our web app and API. Our mission has attracted dozens of researchers who contribute their work for testing and tuning, and we’ve come up with an interesting business model for working together: when their models meet our baseline scores, we provide a revenue share for as long as they continue to perform on our platform. (If you’re interested in participating, we’d love to hear from you!) We have continued to scale our web app and launched an API that we are rolling out to pilot customers. Currently the most popular use cases are: KYC onboarding fraud detection and voice fraud detection (ie. banks, marketplaces); and user-generated deepfake content moderation (ie. social media, dating platforms, news and government organizations). We are currently testing a monthly subscription to scan a minimum of 250 media assets per month. We offer a 30 day pilot that converts into a monthly subscription. If you’d like to give it a try, go to www.realitydefender.ai, click “Request Trial Access” and mention HN in the comments field. We’re here to answer your questions and hear your ideas, and would love to discuss any interesting use cases. We’d also be thrilled to collaborate with anyone who wants to integrate our API or who is working, or would like to work, in this space. We look forward to your comments and conversation!
1 by bpcrd | 0 comments on Hacker News.
Hi HN, we’re Ben, Gaurav and Ali from Reality Defender ( https://ift.tt/4krQmD8 ). We help companies, governments, and journalists determine if media is real or fake, focusing on audio, video and image manipulation. Our API and web app provide real-time scanning, risk scoring, and PDF report cards. Recent advancements in machine learning make it possible to create images, videos and audio of real people saying and doing things they never said or did. The recent spread of this technology has enabled anyone to create highly realistic deepfakes. Although some deepfakes are detectable to the eye by experienced observers who look closely, many people either don’t have experience or are not always looking closely—and of course the technology is only continuing to improve. This marks a leap in the ability of bad actors to distort reality, jeopardizing financial transactions, personal and brand reputations, public opinion, and even national security. We are a team with PhD and Master degrees from Harvard, NYU and UCLA in data science. Between us, we have decades of experience at Goldman Sachs, Google, CIA, FDIC, Dept of Defense and Harvard University Applied Research at the intersection of machine learning and cybersecurity. But our current work began with a rather unlikely project: we tried to duplicate Deepak Chopra. We were working with him to build a realistic deepfake that would allow users to have a real-time conversation with “Digital Deepak” from their iPhones. Creating the Deepak deepfake was surprisingly simple and the result was so alarmingly realistic that we immediately began looking for models that could help users tell a synthetic version from the real thing. We did not find a reliable solution. Frustrated that we’d already spent a week on something we thought would take our coffee break, we doubled down and set out to build our own model that could detect manipulated media. After investigating, we learned why a consistently accurate solution didn’t exist. Companies (including Facebook and Microsoft) were trying to build their own silver-bullet, single-model detection methods—or, as we call it, "one model to rule them all." In our view, this approach will not work because adversaries and the underlying technologies are constantly evolving. For this same reason there will never be a single model to solve anti-virus, malware, etc. We believe that any serious solution to this problem requires a “multi-model'' approach that integrates the best deepfake detection algorithms into an aggregate "model of models." So we trained an ensemble of deep-learning detection models, each of which focuses on its own feature, and then combined the scores. We challenged ourselves to build a scalable solution that integrates the best of our deepfake detection models with models from our collaborators (Microsoft, UC Berkeley, Harvard). We began with a web app proof of concept, and quickly received hundreds of requests for access from governments, companies, and researchers. Our first users turned to our platform for some deepfake scenarios ranging from bad to outright scary: Russian disinformation directed at Ukraine and the West; audio mimicking a bank executive requesting a wire transfer; video of Malaysia’s government leadership behaving scandalously; pornography where participants make themselves appear younger; dating profiles with AI-generated pro pics. All of these, needless to say, are completely fake! As with computer viruses, deepfakes will continue evolving to circumvent current security measures. New deepfake detection techniques must be as iterative as the generation methods. Our solution not only accepts that, but embraces it. We quickly onboard, test, and tune third party models for integration into our model stack, where they can then be accessed via our web app and API. Our mission has attracted dozens of researchers who contribute their work for testing and tuning, and we’ve come up with an interesting business model for working together: when their models meet our baseline scores, we provide a revenue share for as long as they continue to perform on our platform. (If you’re interested in participating, we’d love to hear from you!) We have continued to scale our web app and launched an API that we are rolling out to pilot customers. Currently the most popular use cases are: KYC onboarding fraud detection and voice fraud detection (ie. banks, marketplaces); and user-generated deepfake content moderation (ie. social media, dating platforms, news and government organizations). We are currently testing a monthly subscription to scan a minimum of 250 media assets per month. We offer a 30 day pilot that converts into a monthly subscription. If you’d like to give it a try, go to www.realitydefender.ai, click “Request Trial Access” and mention HN in the comments field. We’re here to answer your questions and hear your ideas, and would love to discuss any interesting use cases. We’d also be thrilled to collaborate with anyone who wants to integrate our API or who is working, or would like to work, in this space. We look forward to your comments and conversation!
New top story on Hacker News: Ask HN: Do you contribute to open source projects?
Ask HN: Do you contribute to open source projects?
6 by dirtylowprofile | 8 comments on Hacker News.
I'm looking to contribute to open source projects but just could not find the time since coding without pay is so foreign to me. I could not just approach my employer and ask if I could help code this open source project. What is your motivation for doing such things, thanks to all the open source maintainers!
6 by dirtylowprofile | 8 comments on Hacker News.
I'm looking to contribute to open source projects but just could not find the time since coding without pay is so foreign to me. I could not just approach my employer and ask if I could help code this open source project. What is your motivation for doing such things, thanks to all the open source maintainers!
Aleksei Navalny, Fiery Putin Critic, Given Additional 9-Year Prison Sentence
By BY ANTON TROIANOVSKI AND VALERIYA SAFRONOVA from NYT World https://ift.tt/bxvr3Gk
Monday, March 21, 2022
Why a California Congressman Has Proposed a Four-Day Workweek
By BY LIVIA ALBECK-RIPKA from NYT U.S. https://ift.tt/c0FwJx7
Market watchers are looking for signs of a slowdown in Russian oil shipments.
By BY STEPHEN GANDEL from NYT Business https://ift.tt/Wz4cfZO
Sunday, March 20, 2022
As Russian forces blaze their way into Mariupol, its mayor warns of forced deportations.
By BY VALERIE HOPKINS from NYT World https://ift.tt/2Fsx4KR
After the ‘Trailer,’ the Blue Jays Are Excited About Their ‘Movie’
By BY JAMES WAGNER from NYT Sports https://ift.tt/wzW6IQA
Russia Will Remake Itself. But It Has to Crumble First.
By BY VARIA BORTSOVA from NYT Opinion https://ift.tt/6EU9Mrp
Saturday, March 19, 2022
Putin Isn’t Yet Ready for Talks With Zelensky, Turkish Official Says
By BY STEVEN ERLANGER from NYT World https://ift.tt/tsE8Uml
Subscribe to:
Posts (Atom)
-
By EMILY COCHRANE and ALAN BLINDER from NYT U.S. https://ift.tt/2RdeOe0
-
Robinhood Is Set to Raise at Least $200 Million in New Funding 139 by jason_zig | 150 comments on Hacker News.
