In this post, we will implement a practical attack on synthetic data models that was described in the Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks by Nicholas Carlini et. al. We will use this attack to see how synthetic data models with various neural network and differential privacy parameter settings actually work at protecting sensitive data and secrets in datasets. And there are some pretty surprising results. Read More
Monthly Archives: April 2021
Forbes AI 50: America’s most promising companies to watch 2021
The Covid-19 pandemic was devastating for many industries, but it only accelerated the use of artificial intelligence across the U.S. economy. Amid the crisis, companies scrambled to create new services for remote workers and students, beef up online shopping and dining options, make customer call centers more efficient and speed development of important new drugs.
Even as applications of machine learning and perception platforms become commonplace, a thick layer of hype and fuzzy jargon clings to AI-enabled software.That makes it tough to identify the most compelling companies in the space—especially those finding new ways to use AI that create value by making humans more efficient, not redundant.
With this in mind, Forbes has partnered with venture firms Sequoia Capital and Meritech Capital to create our third annual AI 50, a list of private, promising North American companies that are using artificial intelligence in ways that are fundamental to their operations. Read More
Tipping the scales in AI: How leaders capture exponential returns
Where many companies tire of marginal gains from early AI efforts, the most successful recognize that the real breakthroughs in AI learning and scale come from persisting through the arduous phases.
Patience is a bitter plant, but its fruit is sweet. This Chinese proverb could well apply to the task of harvesting benefits from artificial intelligence (AI). Many organizations underestimate what it takes to sow true gains, be it selecting the right seeds, apportioning the right investment, or having a mindset willing to put up with the vagaries of the crop cycle. But for those that persevere, the rewards can be huge. McKinsey research finds that leading organizations that approach the AI journey in the right ways and stick with it through the tough patches generate three to four times higher returns from their investments.
These AI leaders get on a different performance trajectory from the outset because they understand that AI is about mastering the long haul. They prepare for that journey by anticipating the types of things that will make it easier to navigate the ups and downs, such as feedback loops that allow data quality and user adoption to compound and AI investments to become self-boosting. Where some companies tire of marginal gains from weeks of effort, leaders recognize that the real breakthroughs in AI learning and scale come from working through those small steps.
But only a small number of businesses have figured out how to make AI work in these ways. Read More
How Pixar’s Movement Animation Became So Realistic | Movies Insider
4 reasons to learn machine learning with JavaScript
In the past few years, Python has become the preferred programming language for machine learning and deep learning. Most books and online courses on machine learning and deep learning either feature Python exclusively or along with R. Python has become very popular because of its rich roster of machine learning and deep learning libraries, optimized implementation, scalability, and versatile features.
But Python is not the only option for programming machine learning applications. There’s a growing community of developers who are using JavaScript to run machine learning models.
While JavaScript is not a replacement for the rich Python machine learning landscape (yet), there are several good reasons to have JavaScript machine learning skills. Here are four. Read More
This has just become a big week for AI regulation
It’s a bumper week for government pushback on the misuse of artificial intelligence.
Today the EU released its long-awaited set of AI regulations, an early draft of which leaked last week. The regulations are wide ranging, with restrictions on mass surveillance and the use of AI to manipulate people.
But a statement of intent from the US Federal Trade Commission, outlined in a short blog post by staff lawyer Elisa Jillson on April 19, may have more teeth in the immediate future. According to the post, the FTC plans to go after companies using and selling biased algorithms. Read More
Microsoft details the latest developments in machine learning at GTC 21
With the rapid pace of change taking place in AI and machine learning technology, it’s no surprise Microsoft had its usual strong presence at this year’s Nvidia GTC event.
Representatives of the company shared their latest machine learning innovations in multiple sessions, covering inferencing at scale, a new capability to train machine learning models across hybrid environments, and the debut of the new PyTorch Profiler that will help data scientists be more efficient when they’re analyzing and troubleshooting ML performance issues.
In all three cases, Microsoft has paired its own technologies, like Azure, with open source tools and NVIDIA’s GPU hardware and technologies to create these powerful new innovations. Read More
How Transformers work in deep learning and NLP: an intuitive introduction
The famous paper “Attention is all you need” in 2017 changed the way we were thinking about attention. With enough data, matrix multiplications, linear layers, and layer normalization we can perform state-of-the-art-machine-translation.
Nonetheless, 2020 was definitely the year of transformers! From natural language now they are into computer vision tasks. How did we go from attention to self-attention? Why does the transformer work so damn well? What are the critical components for its success?
Read on and find out!
A.I. reveals the hidden author of a crucial Bible text
Rescued from the dusty interior of the Qumran Caves in 1947, the Dead Sea Scrolls contain the oldest manuscripts of the Old Testament and are a crucial piece of Biblical history that dates back to the 4th century BCE.
But despite these scrolls’ status as an unmovable piece of religious history, there are still many things that scholars don’t really know about their origin. For example, who actually wrote them down?More like thisInnovation4.16.2021 9:00 AMNASA’s InSight crisis reveals the most difficult part of exploring MarsBy Dave GershgornInnovation4.18.2021 8:00 AMRobotic lawnmowers could cut a huge swath in air pollutionBy Sarah WellsInnovation4.11.2021 8:00 AMCreepy robot skin answers 3 questions about the futureBy Sarah WellsEARN REWARDS & LEARN SOMETHING NEW EVERY DAY.
Using artificial intelligence and pattern recognition, a team of paleographers (scientists who study ancient handwriting) and computer scientists from the University of Groningen have now discovered hidden details in these scrolls that point toward not just one scribe, but two original scribes.
The research was published Wednesday in the journal PLOS One. Read More
How Robust are Randomized Smoothing based Defenses to Data Poisoning?
Predictions of certifiably robust classifiers remain constant in a neighborhood of a point, making them resilient to test-time attacks with a guarantee. In this work, we present a previously unrecognized threat to robust machine learning models that highlights the importance of training-data quality in achieving high certified adversarial robustness. Specifically, we propose a novel bilevel optimization based data poisoning attack that degrades the robustness guarantees of certifiably robust classifiers. Unlike other poisoning attacks that reduce the accuracy of the poisoned models on a small set of target points, our attack reduces the average certified radius(ACR) of an entire target class in the dataset. Moreover, our attack is effective even when the victim trains the models from scratch using state-of-the-art robust training methods such as Gaussian data augmentation[8], MACER[36], and SmoothAdv[29] that achieve high certified adversarial robustness. To make the attack harder to detect, we use clean-label poisoning points with imperceptible distortions. The effectiveness of the proposed method is evaluated by poisoning MNIST and CIFAR10 datasets and training deep neural networks using previously mentioned training methods and certifying the robustness with randomized smoothing. The ACR of the target class, for models trained on generated poi-son data, can be reduced by more than 30%. Moreover, the poisoned data is transferable to models trained with different training methods and models with different architectures. Read More
Rick's Cafe AI 