technology:machine learning algorithm

The “Drunk Pelosi” video shows that cheapfakes can be as damaging as deepfakes.
▻https://slate.com/technology/2019/06/drunk-pelosi-deepfakes-cheapfakes-artificial-intelligence-disinformation.html

https://compote.slate.com/images/31cc8ff6-68d9-4793-a87e-91f16768898e.jpeg?width=780&height=520&rect=1560x1040&offset=0x0

The A.I.-generated “deepfake” video implicitly but unmistakably calls for Facebook to make a public statement on its content moderation polices. The platform has long been criticized for permitting the spread of disinformation and harassment, but it became particularly acute recently, when the company said that it would not remove the “Drunk Pelosi” video.

On Thursday, the House Permanent Select Committee on Intelligence will hold an open hearing on A.I. and the potential threat of deepfake technology to Americans. Many technology researchers believe that deepfakes—realistic-looking content developed using machine learning algorithms—will herald a new era of information warfare. But as the “Drunk Pelosi” video shows, slight edits of original videos may be even more difficult to detect and debunk, creating a cascade of benefits for those willing to use these digital dirty tricks.

The video, posted to a self-described news Facebook page with a fan base of about 35,000, depicted Nancy Pelosi slurring her words and sounding intoxicated. However, when compared with another video from the same event, it was clear even to nonexperts that it had been slowed down to produce the “drunken” effect. Call it a “cheapfake”—it was modified only very slightly. While the altered video garnered some significant views on Facebook, it was only after it was amplified by President Donald Trump and other prominent Republicans on Twitter that it became a newsworthy issue. The heightened drama surrounding this video raises interesting questions not only about platform accountability but also about how to spot disinformation in the wild.

“Cheapfakes” rely on free software that allows manipulation through easy conventional editing techniques like speeding, slowing, and cutting, as well as nontechnical manipulations like restaging or recontextualizing existing footage that are already causing problems. Cheapfakes already call into question the methods of evidence that scientists, courts, and newsrooms traditionally use to call for accountability

Many will never know the video was a fake, but the advantages it gave to pundits will echo into the future. It’s a recent example of what legal theorists Bobby Chesney and Danielle Citron call the liar’s dividend . Those wishing to deny the truth can create disinformation to support their lie, while those caught behaving badly can write off the evidence of bad behavior as disinformation. In a new survey from Pew Research Center, 63 percent of respondents said that they believe altered video and images are a significant source of confusion when it comes to interpreting news quality. That loss of trust works in favor of those willing to lie, defame, and harass to gain attention.

As Daniel Kreiss and others have pointed out, people don’t just share content because they believe it. They do it for a host of reasons, not the least of which is simply because a message speaks to what users see as an implicit truth of the world even as they know it is not factually true. Researchers have found that creating and sharing hateful, false, or faked content is often rewarded on platforms like Facebook.

The looming threat of the deepfake is worth attention—from politicians, like at the upcoming hearing; from journalists; from researchers; and especially from the public that will ultimately be the audience for these things. But make no mistake: Disinformation doesn’t have to be high tech to cause serious damage.

#Fake_news #Deep_fake #Cheap_fake #Nancy_Pelosi #Médias_sociaux

12 Key Lessons from ML researchers and practitioners
▻https://hackernoon.com/12-key-lessons-from-ml-researchers-and-practitioners-3d4818a2feff?source

Machine learning algorithms come with the promise of being able to figure out how to perform important tasks by learning from data, i.e., generalizing from examples without being explicitly told what to do. This means that the higher the amount of data, the more ambtious problems can be tackled by these algorithms. However, developing successful machine learning applications requires quite some “black art” that is hard to find in text books or introductory courses on machine learning.I recently stumbled upon a great research paper by Professor Pedro Domingos that puts together lessons learned by machine learning researchers and practitioners. In this post, I am going to walkthrough those lessons with you.Get ready to learn about: pitfalls to avoid, important issues to focus on, and (...)

#machine-learning #development #best-practices #artificial-intelligence #data-science

Even Smarter: Achieving #aiops in the Age of Big Data
▻https://hackernoon.com/even-smarter-achieving-aiops-in-the-age-of-big-data-8240be07208?source=r

https://cdn-images-1.medium.com/max/1024/1*FrYXHL-1eSKqsMP13nUAdg.jpeg

How is a new era of data challenges changing operation and maintenance for IT? Best practice form the Alibaba tech team.This article is part of the AIOps for Big Data series.The heart of artificial intelligence is data, and companies with a lot of it are constantly working to boost what it can do for them. Beyond that simple fact, though, many continue to wonder what it means to say that data is involved in development practices like AIOps (AI for IT operations), or how data can be used in place of human and even machine-driven analytics.Further questions abound. How can we use machine learning algorithms together with big data-based business operation and maintenance platforms? How can machine learning enhance alarm filtering, anomaly monitoring, automated repairs, and other tasks to (...)

#machine-learning #artificial-intelligence #big-data #big-data-aiops

How to make #snapchat Lenses ?
▻https://hackernoon.com/how-to-make-snapchat-lenses-f9eae861b5db?source=rss----3a8144eabfe3---4

https://cdn-images-1.medium.com/max/600/1*yCyEOoUBa-oYYT9xyr1mKA.gif

We all love SnapChat lenses/filters, but ever wondered how you can make your own? This article explains how you can use python and computer vision libraries like opencv and dlib to create your own “glasses and mustache lens” with as few as 80 lines of code.What is dlib?Dlib is a modern C++ toolkit containing machine learning algorithms and tools for creating complex software in C++ to solve real world problems. It is used in both industry and academia in a wide range of domains including robotics, embedded devices, mobile phones, and large high performance computing environments. Dlib’s open source licensing allows you to use it in any application, free of charge. [source]▻https://medium.com/media/66cfd3663b00a2d2c3ded8a04e8c8bb5/hrefWhat is OpenCV?OpenCV (Open Source Computer Vision) is a (...)

#computer-vision #snapchat-lenses #machine-learning #make-snapchat-lense

Boosting and Bagging: How To Develop A Robust Machine Learning Algorithm
▻https://hackernoon.com/how-to-develop-a-robust-algorithm-c38e08f32201?source=rss----3a8144eabfe

https://cdn-images-1.medium.com/max/1024/1*hNnC9r25vEAuRffNy2mjdw.jpeg

Bootstrapping/Bagging/BoostingMachine learning and data science require more than just throwing data into a python library and utilizing whatever comes out.Data scientists need to actually understand the data and the processes behind the data to be able to implement a successful system.One key methodology to implementation is knowing when a model might benefit from utilizing bootstrapping methods. These are what are called ensemble models. Some examples of ensemble models are AdaBoost and Stochastic Gradient Boosting.Why use ensemble models?They can help improve algorithm accuracy or improve the robustness of a model. Two examples of this is boosting and bagging. Boosting and Bagging are must know topics for data scientists and machine learning engineers. Especially if you are planning (...)

#machine-learning #statistics #data-science #algorithms #self-improvement

Choosing the Right Machine Learning Algorithm
▻https://hackernoon.com/choosing-the-right-machine-learning-algorithm-68126944ce1f?source=rss---

https://cdn-images-1.medium.com/max/720/1*Xf23cuffXwWbTzetaFiC8w.png

Machine learning is part art and part science. When you look at machine learning #algorithms, there is no one solution or one approach that fits all. There are several factors that can affect your decision to choose a machine learning algorithm.Some problems are very specific and require a unique approach. E.g. if you look at a recommender system, it’s a very common type of machine learning algorithm and it solves a very specific kind of problem. While some other problems are very open and need a trial & error approach. Supervised learning, classification and regression etc. are very open. They could be used in anomaly detection, or they could be used to build more general sorts of predictive models.Besides some of the decisions that we make when choosing a machine learning (...)

#machine-learning #ml-algorithm #learning-algorithms #data-science

Introduction to Machine Learning Algorithms: Logistic Regression
▻https://hackernoon.com/introduction-to-machine-learning-algorithms-logistic-regression-cbdd82d8

https://cdn-images-1.medium.com/max/1024/1*SMlRX2-yBaV-1LBt8Dl7rg.jpeg

Build your Logistic Regression model from scratchLogistic regression is the most famous machine learning algorithm after linear regression. In a lot of ways, linear regression and logistic regression are similar. But, the biggest difference lies in what they are used for. Linear regression algorithms are used to predict/forecast values but logistic regression is used for classification tasks. If you are shaky on the concepts of linear regression, check this out. There are many classification tasks done routinely by people. For example, classifying whether an email is a spam or not, classifying whether a tumour is malignant or benign, classifying whether a website is fraudulent or not, etc. These are typical examples where machine learning algorithms can make our lives a lot easier. A (...)

#code #tutorial #logistic-regression #machine-learning

This Neural Net Hallucinates Sheep - Facts So Romantic
▻http://nautil.us/blog/this-neural-net-hallucinates-sheep

http://static.nautil.us/14298_56de534cf589f6d0ca09630db803ba00.jpg

If you’ve been on the internet today, you’ve probably interacted with a neural network. They’re a type of machine learning algorithm that’s used for everything from language translation to finance modeling. One of their specialties is image recognition. Several companies—including Google, Microsoft, IBM, and Facebook—have their own algorithms for labeling photos. But image recognition algorithms can make really bizarre mistakes.Janelle ShaneMicrosoft Azure’s computer vision API added the above caption and tags. But there are no sheep in the image of above. None. I zoomed all the way in and inspected every speck. Janelle ShaneIt also tagged sheep in this image. I happen to know there were sheep nearby. But none actually present.It hasn’t realized that “sheep” means the actual animal, not just a (...)

The most detailed maps of the world will be for cars, not humans | Ars Technica
▻https://arstechnica.com/cars/2017/03/the-most-detailed-maps-of-the-world-will-be-for-cars-not-humans

https://cdn.arstechnica.net/wp-content/uploads/2017/03/HD-Live-Map-980x663.png

But a more scalable solution involves leveraging the embedded sensors in cars already using HD maps to navigate. “Today, many high-end vehicles not only have LTE connections but an array of advanced sensors on them; a forward-looking camera, forward and rear radar, and we’re working with numerous OEMs on getting that sensor data transmitted into our cloud,” said Sood. That’s going to mean millions of data points being constantly uploaded to the cloud, where Here’s servers will do the heavy lifting of making sense out of it all.

“We have machine learning algorithms and aggregation algorithms that sift through that data and try distilling changes from noise or aberrations in the world. Once we’ve detected those changes we update the map and then deploy that to the vehicle,” explained Sood.

#cartographie #autopilote #voiture_autopilotée #machine-learning #robotisation

Can AI and big data improve how you get news? Cheetah Mobile is making a $57M bet that it can - TechRepublic
▻http://www.techrepublic.com/article/can-ai-and-big-data-improve-how-you-get-news-cheetah-mobile-is-making-

On Friday, one of China’s leading tech companies, Cheetah Mobile, announced the $57 million acquisition of New Republic—a move that signals its “journey of a transformation,” according to CTO Charles Fan, highlighting its investment in AI and transition into a mobile content company.

Founded in 2010, Cheetah Mobile began as a mobile tools provider. Over the last six years, it has become a major player in China’s tech scene—and made a big impact, globally. It’s the second largest internet and mobile security company in China, and the third global non-gaming developer on the Google Play Store, providing the popular apps Piano Tiles 2, Clean Master, CM Security, and Battery Doctor. Fan told TechRepublic that the company has over 650,000,000 monthly active users, internationally, in Q1.

Fan said he sees the company as a “bridge between China and the world.” What makes Cheetah Mobile different, he said, is that 80% of their mobile users are outside of China. “Even though we are headquartered in Beijing, our target audience is mostly overseas,” he said. While most companies are based solely in China, Cheetah Mobile recently set up a second headquarters in Silicon Valley, expanding its global footprint and assisting in the development of its machine learning algorithms.
[…]
“Our goal is to grow News Republic into being the smartest news app,” Fan said, “so that users can get the content they want to consume, plus help content producers, game developers, and influencers.”

And, for the huge amount of data Cheetah Mobile is collecting on the phones? It’s helping the algorithms become even smarter.

A Visual Search Engine for the Aerial Patterns of Cities

▻http://hyperallergic.com/301858/a-visual-search-engine-for-the-aerial-patterns-of-cities

Terrapattern, developed at the Carnegie Mellon Frank-Ratchye STUDIO for Creative Inquiry, is the first open-access visual search tool for satellite imagery. It is currently available for Pittsburgh, San Francisco, New York City, Detroit, Austin, Miami, and Berlin. This means you may scan these cities’ landscapes for common forms of your particular interest that are not conventionally labelled on a map: circular backyard pools or cul-de-sacs, perhaps, or even dilapidated nautical wrecks. All you have to do is find the tile of topography that intrigues you, and dozens of search results of similar views will arrive courtesy of machine learning algorithms trained to sift through images from OpenStreetMap. You can then export these images as a geographic text file.

http://hyperallergic.com/wp-content/uploads/2016/05/pools-in-pittsburgh.png http://hyperallergic.com/wp-content/uploads/2016/05/tennis.jpg http://hyperallergic.com/wp-content/uploads/2016/05/detroit-768x512.jpg http://hyperallergic.com/wp-content/uploads/2016/05/waves.jpg http://hyperallergic.com/wp-content/uploads/2016/05/culdesacs.jpg

The NSA’s SKYNET program may be killing thousands of innocent people | Ars Technica UK
►http://arstechnica.co.uk/security/2016/02/the-nsas-skynet-program-may-be-killing-thousands-of-innocent-people

Last year, The Intercept published documents detailing the NSA’s SKYNET programme. According to the documents, #SKYNET engages in mass surveillance of Pakistan’s mobile phone network, and then uses a machine learning algorithm on the cellular network metadata of 55 million people to try and rate each person’s likelihood of being a terrorist.

Patrick Ball—a data scientist and the director of research at the Human Rights Data Analysis Group—who has previously given expert testimony before war crimes tribunals, described the NSA’s methods as “ridiculously optimistic” and “completely bullshit.” A flaw in how the #NSA trains SKYNET’s machine learning algorithm to analyse cellular metadata, Ball told Ars, makes the results scientifically unsound.

Somewhere between 2,500 and 4,000 people have been killed by #drone strikes in Pakistan since 2004, and most of them were classified by the US government as “extremists,” the Bureau of Investigative Journalism reported. Based on the classification date of “20070108” on one of the SKYNET slide decks (which themselves appear to date from 2011 and 2012), the machine learning program may have been in development as early as 2007.

The NSA’s SKYNET program may be killing thousands of innocent people | Ars Technica UK
►http://arstechnica.co.uk/security/2016/02/the-nsas-skynet-program-may-be-killing-thousands-of-innocent-people

“Ridiculously optimistic” machine learning algorithm is “completely bullshit,” says expert.

In 2014, the former director of both the CIA and NSA proclaimed that “we kill people based on metadata.” Now, a new examination of previously published Snowden documents suggests that many of those people may have been innocent.

http://cdn.arstechnica.net/wp-content/uploads/sites/3/2016/02/skynet-courier-detection-via-machine-learning-p3-normal-640x495.gif

(...) The highest scoring selector who travelled to Peshawar and Lahore is “PROB AHMED ZAIDAN”, Al-Jazeera’s long-time bureau chief in Islamabad.
As The Intercept reported, Zaidan frequently travels to regions with known terrorist activity in order to interview insurgents and report the news. But rather than questioning the machine learning that produced such a bizarre result, the NSA engineers behind the algorithm instead trumpeted Zaidan as an example of a SKYNET success in their in-house presentation, including a slide that labelled Zaidan as a “MEMBER OF AL-QA’IDA.”

http://cdn.arstechnica.net/wp-content/uploads/sites/3/2016/02/skynet-courier-detection-via-machine-learning-p17-normal-640x495.gif

#machine_learning #stupidité_artificielle #drones #robots_tueurs

How One Intelligent Machine Learned to Recognize Human Emotions | MIT Technology Review
▻http://www.technologyreview.com/view/545986/how-one-intelligent-machine-learned-to-recognize-human-emotions

Nobody knew how to identify people’s emotional states by looking at their brain waves. Then a machine learning algorithm stepped in.

▻https://d267cvn3rvuq91.cloudfront.net/i/images/Emotional%20states.PNG?sw=790

#machine_learning #émotions #recherche #chine #it_has_begun