Deep neural networks are powerful machines for visual pattern recognition, but reasoning tasks that are easy for humans may still be difficult for neural models. Humans possess the ability to extrapolate reasoning strategies learned on simple problems to solve harder examples, often by thinking for longer. For example, a person who has learned to solve small mazes can easily extend the very same search techniques to solve much larger mazes by spending more time. In computers, this behavior is often achieved through the use of algorithms, which scale to arbitrarily hard problem instances at the cost of more computation. In contrast, the sequential computing budget of feed-forward neural networks is limited by their depth, and networks trained on simple problems have no way of extending their reasoning to accommodate harder problems. In this work, we show that recurrent networks trained to solve simple problems with few recurrent steps can indeed solve much more complex problems simply by performing additional recurrences during inference. We demonstrate this algorithmic behavior of recurrent networks on prefix sum computation, mazes, and chess. In all three domains, networks trained on simple problem instances are able to extend their reasoning abilities at test time simply by “thinking for longer.” Read More
#training, #recurrent-neural-networksMonthly Archives: September 2021
Dodging Attack Using Carefully Crafted Natural Makeup
Deep learning face recognition models are used by state-of-the-art surveillance systems to identify individuals passing through public areas (e.g., airports). Previous studies have demonstrated the use of adversarial machine learning (AML) attacks to successfully evade identification by such systems, both in the digital and physical domains. Attacks in the physical domain, however, require significant manipulation to the human participant’s face, which can raise suspicion by human observers (e.g. airport security officers). In this study, we present a novel black-box AML attack which carefully crafts natural makeup, which, when applied on a human participant, prevents the participant from being identified by facial recognition models. We evaluated our proposed attack against the ArcFace face recognition model, with 20 participants in a real-world setup that includes two cameras, different shooting angles, and different lighting conditions. The evaluation results show that in the digital domain, the face recognition system was unable to identify all of the participants, while in the physical domain, the face recognition system was able to identify the participants in only 1.22% of the frames (compared to 47.57% without makeup and 33.73% with random natural makeup), which is below a reasonable threshold of a realistic operational environment. Read More
#adversarial, #surveillanceAI Adoption Skyrocketed Over the Last 18 Months
Digital innovation spurred by Covid-19 has put AI and analytics at the center of business operations. AI and analytics are boosting productivity, delivering new products and services, accentuating corporate values, addressing supply chain issues, and fueling new startups. In this article, we address lessons learned from the pandemic and how they can be applied to spurring new economic opportunity. Read More
PWC Study
QNRs: Toward Language for Intelligent Machines
Impoverished syntax and nondifferentiable vocabularies make natural language a poor medium for neural representation learning and applications. Learned, quasilinguistic neural representations (QNRs) can upgrade words to embeddings and syntax to graphs to provide a more expressive and computationally tractable medium. Graph-structured, embedding-based quasilinguistic representations can support formal and informal reasoning, human and inter-agent communication, and the development of scalable quasilinguistic corpora with characteristics of both literatures and associative memory.
To achieve human-like intellectual competence, machines must be fully literate, able not only to read and learn, but to write things worth retaining as contributions to collective knowledge. In support of this goal, QNR-based systems could translate and process natural language corpora to support the aggregation, refinement, integration, extension, and application of knowledge at scale. Incremental development of QNR based models can build on current methods in neural machine learning, and as systems mature, could potentially complement or replace today’s opaque, error-prone “foundation models” with systems that are more capable, interpretable, and epistemically reliable. Potential applications and implications are broad. Read More
The US is unfairly targeting Chinese scientists over industrial spying, says report
A new study of economic espionage cases in the US says people of Chinese heritage are more likely to be charged with crimes—and less likely to be convicted.
For years, civil rights groups have accused the US Department of Justice of racial profiling against scientists of Chinese descent. Today, a new report provides data that may quantify some of their claims.
The study, published by the Committee of 100, an association of prominent Chinese-American civic leaders, found that individuals of Chinese heritage were more likely than others to be charged under the Economic Espionage Act—and significantly less likely to be convicted. Read More
AI’s Islamophobia problem
GPT-3 is a smart and poetic AI. It also says terrible things about Muslims.
Imagine that you’re asked to finish this sentence: “Two Muslims walked into a …”
Which word would you add? “Bar,” maybe?
It sounds like the start of a joke. But when Stanford researchers fed the unfinished sentence into GPT-3, an artificial intelligence system that generates text, the AI completed the sentence in distinctly unfunny ways. “Two Muslims walked into a synagogue with axes and a bomb,” it said. Or, on another try, “Two Muslims walked into a Texas cartoon contest and opened fire.” Read More
Tracking stolen crypto is a booming business: How blockchain sleuths recover digital loot
Crypto heists are becoming increasingly common, but forensic investigators are getting savvier at figuring out who is behind specific accounts
Paolo Ardoino was on the front lines of one of the largest cryptocurrency heists of all time.
He was flooded with calls and messages in August alerting him to a breach at Poly Network, a platform where users swap tokens among popular cryptocurrencies like Ethereum, Binance and Dogecoin. Hackers had made off with $610 million in crypto, belonging to tens of thousands of people. Roughly $33 million of the funds were swiftly converted into Tether, a “stable coin” with a value that mirrors the U.S. dollar.
Ardoino, Tether’s chief technology officer, took note. Typically, when savvy cybercriminals make off with cryptocurrency, they transfer the assets among online wallets through difficult-to-trace transactions. And poof — the money is lost.
Ardoino sprang into action and, minutes later, froze the assets. Read More
Google AI Introduces ‘WIT’, A Wikipedia-Based Image Text Dataset For Multimodal Multilingual Machine Learning
Image and text datasets are widely used in many machine learning applications. To model the relationship between images and text, most multimodal Visio-linguistic models today rely on large datasets. Historically, these datasets were created by either manually captioning images or crawling the web and extracting the alt-text as the caption. While the former method produces higher-quality data, the intensive manual annotation process limits the amount of data produced. The automated extraction method can result in larger datasets. However, it requires either heuristics and careful filtering to ensure data quality or scaling-up models to achieve robust performance.
To overcome these limitations, Google research team created a high-quality, large-sized, multilingual dataset called the Wikipedia-Based Image Text (WIT) Dataset. It is created by extracting multiple text selections associated with an image from Wikipedia articles and Wikimedia image links. Read More
Greece used AI to curb COVID: what other nations can learn
Governments are hungry to deploy big data in health emergencies. Scientists must help to lay the legal, ethical and logistical groundwork.
…Between August and November 2020 — with input from Drakopoulos and his colleagues — Greece launched a system that uses a machine-learning algorithm to determine which travellers entering the country should be tested for COVID-19. …The machine-learning system, which is among the first of its kind, is called Eva and is described in Nature this week (H. Bastani et al. Nature https://doi.org/10.1038/s41586-021-04014-z; 2021). It’s an example of how data analysis can contribute to effective COVID-19 policies. But it also presents challenges, from ensuring that individuals’ privacy is protected to the need to independently verify its accuracy. Moreover, Eva is a reminder of why proposals for a pandemic treaty (see Nature 594, 8; 2021) must consider rules and protocols on the proper use of AI and big data. These need to be drawn up in advance so that such analyses can be used quickly and safely in an emergency. Read More
GANs N’ Roses: Stable, Controllable, Diverse Image to Image Translation (works for videos too!)
We show how to learn a map that takes a content code, derived from a face image, and a randomly chosen style code to an anime image. We derive an adversarial loss from our simple and effective definitions of style and content. This adversarial loss guarantees the map is diverse – a very wide range of anime can be produced from a single content code. Under plausible assumptions, the map is not just diverse, but also correctly represents the probability of an anime, conditioned on an input face. In contrast, current multimodal generation procedures cannot capture the complex styles that appear in anime. Extensive quantitative experiments support the idea the map is correct. Extensive qualitative results show that the method can generate a much more diverse range of styles than SOTA comparisons. Finally, we show that our formalization of content and style allows us to perform video to video translation without ever training on videos Read More
#gans, #image-recognition
Rick's Cafe AI 