Developed by Elon Musk-owned OpenAI, GPT-3 is the autoregressive language model that deploys deep learning to produce human-like text. OpenAIs GPT-3 is currently the largest artificial intelligence language model, marred in debates that range from whether it is a step closer to AGI (Artificial General Intelligence) or it is the first step toward creating this sort of superintelligence.

GPT-3 (generative pre-trained transformer.) is the third in a series of autocomplete tools designed by OpenAI. GPT-3 program has been trained on a huge corpus of text stored as billions of weighted connections between the different nodes in GPT-3s neural network. The program looks and finds patterns without any guidance, which it then uses to complete text prompts. If you input the word fire into GPT-3, the program knows, based on the weights in its network, that the words alarm and water are much more likely to follow than soil or forests.

GPT-3is trained on 175 billion parameters that are more than 100 times more than its predecessor and ten times more than comparable programs, to complete a mind-boggling array of autocomplete tasks, whose sharpness astonishes mankind!

The dataset GPT-3 was entirety trained on-

The English Wikipedia, spanning some 6 million articles which makes up only 0.6 percent of its training data.

Digitized books and various web links, including news articles, recipes, and poetry, coding manuals, fanfiction, religious prophecy, and whatever else imaginable!

Any type of good and bad text that has been uploaded on the internet including the potentially harmful conspiracy theories, racist screeds, pseudoscientific textbooks, and the manifestos of mass shooters.

Its hardly comprehensive, but heres a small sample of things people have created with GPT-3:

A chatbot that talks to historical figures

Because GPT-3 has been trained on so many digitized books, it has assimilated a fair amount of knowledge relevant to specific thinkers. Leverage GPT-3 to make a chatbot talk like the philosopher Bertrand Russell, and ask him to explain his views. Fictional characters are as accessible to GPT-3 as historical ones. Check out the exciting dialogue between Alan Turing and Claude Shannon, interrupted by Harry Potter!

Makes your own quizzes

Definitely a blessing to the education system, GPT-3 is an awesome helper of teachers as well as students. It will generate Quizzes for practice on any topics and also explain the answers to these questions in detail, helping students to learn anything from anyone be it robotics from Elon Musk, physics from Newton, relativity theory from Einstein, and literature from Shakespeare.

A question-based search engine

Trained on the entire Wikipedia, GPT-3 is like Google but for questions and answers. Type a question and GPT-3 directs you to the relevant Wikipedia URL for the answer.

Answer medical queries

A medical student from the UK used GPT-3 to answer health care questions. The program not only gave the right answer but correctly explained the underlying biological mechanism.

Style transfer for text

The input text is written in a certain style and GPT-3 can change it to another. In an example on Twitter, a user input text in plain language and asked GPT-3 to change it to legal language. This transforms inputs from my landlord didnt maintain the property to The Defendants have permitted the real property to fall into disrepair and have failed to comply with state and local health and safety codes and regulations.

Compose its own Music

Guitar tabs are shared on the web using ASCII text files, which comprise part of GPT-3s training dataset. Naturally, that means GPT-3 can generate music itself after being given a few chords to start.

Write creative fiction

This is a wide-ranging area within GPT-3s skillset but an incredibly impressive one. The best collection of the programs literary samples comes from independent researcher and writer Gwern Branwen who has collected a trove of GPT-3s writing. It ranges from a type of one-sentence pun known as a Tom Swifty to poetry in the style of Allen Ginsberg, T.S. Eliot, and Emily Dickinson to Navy SEAL copypasta.

Autocomplete images, not just text

The basic GPT architecture can be retrained on pixels instead of words, allowing it to perform the same autocomplete tasks with visual data than it does with text input.

Solving language and syntax puzzles

You can show GPT-3 certain linguistic patterns (Like truck driver becomes driver of truck and chocolate cake becomes cake made of chocolate) and it will complete any new prompts you show it correctly. However, being in the nascent stage, a lot of developments are still bound to happen. As computer science professor Yoav Goldberg whos been sharing lots of these examples on Twitter puts it, such abilities are new and super exciting for AI, but they dont mean GPT-3 has mastered language.

Code generation based on text descriptions

Describe a design element or page layout of your choice in simple words and GPT-3 spits out the relevant code. Users have used GPT-3 to generate code for a machine learning model, just by describing the dataset and required output. In another example, in a layout generator, you have to describe any layout you want, and GPT-3 will generate the JSX code for you.

A world of Unlimited Possibilities has just Begun!

Read more from the original source:

Discover Unlimited Possibilities with OpenAI's AI Tool GPT-3 - Analytics Insight

STANFORD, Calif., Aug. 10, 2020 /PRNewswire/ --TheStanford Center for Health Education launched an online program in AI and Healthcare this week. The program aims to advance the delivery of patient care and improve global health outcomes through artificial intelligence and machine learning.

The online program, taught by faculty from Stanford Medicine, is designed for healthcare providers, technology professionals, and computer scientists. The goal is to foster a common understanding of the potential for AI to safely and ethically improve patient care.

Stanford University is a leader in AI research and applications in healthcare, with expertise in health economics, clinical informatics, computer science, medical practice, and ethics.

"Effective use of AI in healthcare requires knowing more than just the algorithms and how they work," said Nigam Shah, associate professor of medicine and biomedical data science, the faculty director of the new program. "Stanford's AI in Healthcare program will equip participants to design solutions that help patients and transform our healthcare system. The program will provide a multifaceted perspective on what it takes to bring AI to the clinic safely, cost-effectively, and ethically."

AI has the potential to enable personalized care and predictive analytics, using patient data. Computer system analyses of large patient data sets can help providers personalize optimal care. And data-driven patient risk assessment canbetter enable physicians to take the right action, at the right time. Participants in the four-course program will learn about: the current state, trends and implications of artificial intelligence in healthcare; the ethics of AI in healthcare; how AI affects patient care safety, quality, and research; how AI relates to the science, practice and business of medicine; practical applications of AI in healthcare; and how to apply the building blocks of AI to innovate patient care and understand emerging technologies.

The Stanford Center for Health Education (SCHE), which created the AI in Healthcare program, develops online education programs to extend Stanford's reach to learners around the world. SCHE aims to shape the future of health and healthcare through the timely sharing of knowledge derived from medical research and advances. By facilitating interdisciplinary collaboration across medicine and technology, and introducing professionals to new disciplines, the AI in Healthcare program is intended to advance the field.

"In keeping with the mission of the Stanford Center for Health Education to expand knowledge and improve health on a global scale, we are excited to launch this online certificate program on Artificial Intelligence in Healthcare," said Dr. Charles G. Prober, founding executive director of SCHE. "This program features several of Stanford's leading thinkers in this emerging field a discipline that will have a profound effect on human health and disease in the 21st century."

The Stanford Center for Health Education is a university-wide program supported by Stanford Medicine. The AI in Healthcare program is available for enrollment through Stanford Online, and hosted on the Coursera online learning platform. The program consists of four online courses, and upon completion, participants can earn a Stanford Online specialization certificate through the Coursera platform. The four courses comprising the AI in Healthcare specialization are: Introduction to Healthcare, Introduction to Clinical Data, Fundamentals of Machine Learning for Healthcare, and Evaluations of AI Applications in Healthcare.

SOURCE Stanford Center for Health Education

Go here to see the original:

Stanford Center for Health Education Launches Online Program in Artificial Intelligence in Healthcare to Improve Patient Outcomes - PRNewswire

America is poised to invest billions of dollars to remain the leader in artificial intelligence as well as quantum computing.

This investment is critically needed to reinvigorate the science that will shape our future. But in order to get the most from this investment, we have to create an environment that will produce innovations that are not just technical advancements but will also benefit society and uplift everybody in our society.

This is why it is important to invest in fixing the systemic inequalities that have sidelined Black people from contributing to AI and from having a hand in the products that will undoubtedly impact everyone. Black scholars, engineers, and entrepreneurs currently have little-to-no voice in AI.

There are a number of bills coming through the House and the Senate to invest up to $100 billion in the fields of AI and quantum computing. This legislation, for example, the one from the House Committee on Science, Space, and Technology, makes references to the importance of ethics, fairness, and transparency, which are great principles but are not precise and lack a clear meaning. The bicameral Endless Frontier Act would effect transformational change to AI but is similarly unclear about how it would remedy institutional inequity in AI and address the lived experience of Black Americans. What these bills do not address is equal opportunity, which has a more precise meaning and is grounded in the movement for civil rights. These substantial investments in technology should help us realize equity and better outcomes in tech research and development. They should ensure that the people building these technologies reflect society. We are not seeing that right now.

As a Black American, I am deeply concerned about the outcomes and ill-effects that this surge of funding could produce if we do not have diversity in our development teams, our research labs, our classrooms, our boardrooms, and our executive suites.

If you look at companies building AI today like OpenAI, Google DeepMind, Clearview, and Amazon they are far from having diverse development teams or diverse executive teams. And we are seeing the result play out in the wrongful AI-triggered arrest of Robert Williams in January, as well as many other abuses that go under the radar.

Thus, we need to see these substantial government investments in AI tied to clear accountability for equal opportunity. If we can bring equal opportunity and technological advancement together, we will deliver the potential of AI in a way that will benefit society as a whole and live up to the ideals of America.

So, how do we ensure equal opportunity in tech development? It starts with how we invest in scientific research. Currently, when we make investments, we only think about technological advancement. Equal opportunity is a non-priority and, at best, a secondary consideration.

This is the entrenched system of innovation that we are used to seeing. Scientific research is the spring-well that fuels advancements in our productivity and quality of life. Science has yielded an incredible return on investment across our history and is continually transforming our lives. But we also need innovation inside our engine of innovation as well. It would be a mistake to assume that all scientists are enlightened enough to engage, train, mentor, cultivate, and include Black people. We should always ask: What is the bottom line that incentivizes and shapes our scientific effort?

The fix is simple really and something we can do almost immediately: We must start enforcing existing civil rights statutes for how government funds are distributed in support of scientific advancement. This will mostly affect universities, but it will also reform other organizations that are leading the way in artificial intelligence.

Think of the government as the venture capitalist that specifically has the interest of the people as its bottom line.

If we start enforcing existing civil right statues, then federal funding of artificial intelligence will create a virtuous cycle. It is not just advanced technology and ideas that come out of that funding. It is also the people produced from supported research labs who are trained in how to engineer and innovate.

And research labs have an impact on the science classrooms. The faculty and students engaged in research are also educating the next generation innovation workforce. They impact not only who is in the classroom environment but also who gets opportunities on the development teams that define the industry. Government funding should remind universities of their responsibility to mentor and grow future generations, not just pick winners and losers by grade policing.

If we fix how we invest in science with this massive influx of money, we can produce more enlightened innovators that will produce better products and AI that will help remedy some of the troubling things we are seeing right now with the technology. We will also be able to produce new technologies that expand our horizons beyond our current imaginations and dogma.

If a research lab or a university degree program is not diverse and not creating equal opportunity as required by law, then it should be ineligible for federal funding, including research grants. We should not fund researchers in computer science departments that have only yielded token representation of Black students in their graduating classes. We should not fund researchers who have received millions in public money but have never successfully mentored a Black student. Instead, we should reward researchers who achieve both inclusion of Black scholars and scientific excellence in their work. We should incentivize thoughtful and considerate mentorship by researchers, as we would want for ourselves, our own children, and our tuition dollars.

We should look at equal opportunity the same way as we look at investing in the stock market. Would you invest in a stock that has not shown any growth that has stagnated and come to perform badly? It is unlikely anybody would put their own money in that stock unless they saw evidence growth will occur. The same should hold true for university departments that build their prestige and economic viability primarily from money granted by the American taxpayer.

Who would be responsible for making these decisions? Ideally, it would be done by federal funding agencies themselves the National Science Foundation, the National Institutes of Health, the Department of Defense, etc. These agencies have yielded an immense return on investment that has enabled American innovation to grow exponentially over the last century, but their view of merit needs to be rethought in the context of 2020 and the realities of our new century.

I wrote earlier that this was an easy fix. And it is, on paper. But change will be difficult for research institutions because of their entrenched institutional culture. The people who are in positions to make the necessary change have come up through the system. And so they do not necessarily see the solution or the problem.

I am a Professor of Computer Science and Engineering at the University of Michigan. I have worked in robotics and artificial intelligence for over 20 years. I know the feelings of elation and validation from winning large federal grants to support my research and my students. Few words can describe the sense of honor and acknowledgment that comes with federal support of ones research. I still swell with pride every time I think about my opportunity to shake President George W. Bushs hand in 2007 and the congratulatory note in 2016 from my congressional representative, Rep. Debbie Dingle, for my National Robotics Initiative grant.

I also understand from experience how hard it is to see things from the inside. If we make the analogy to law enforcement, it is very much like the police policing the police. We are the people that are producing the technology innovation and benefiting from the funding, but we are also responsible for reviewing ourselves. There is little external accountability, with only evolving attempts at broadening participation from within.

I am neither a lawyer nor a member of the civil service, to be very clear. That said, this moment in our history is an opportune time to reimagine equal opportunity throughout the federal research portfolio. One possibility is through the creation of an independent agency that analyzes and enforces equal opportunity across programs for federal funding of scientific research, in contrast to dividing this responsibility among individual sub-agencies solely within the Executive Branch. Regardless of implementation, it is essential that we continually oversee the policies and practices of funding in artificial intelligence to make sure there is proper representation and diversity included and to ensure that our federal funding is not going to be spent without consideration of different viewpoints on how technology should be built, and of the larger systemic issues at play.

The time to act on this is now before the funding begins. When it comes to discrimination and racism, we must address both the hidden disparate impact in our systems of innovation as well as the traditional explicit disparate treatment (such as the vividly portrayed in the 2016 movie Hidden Figures).

For those who want to act, you can first look at your own organization and your own working environments and see whether you are living up to the civil rights statutes. If you are interested in translating protest into policy, write to your representatives in Congress and your elected officials and tell them equal opportunity in AI is important.

We should also ask our presidential candidates to commit to the kind of accountability I have outlined here. Regardless of who is elected, these issues of artificial intelligence and equal opportunity are going to define our country for the next few decades. It is a national priority that demands our attention at the highest levels. We should all be asking who is developing this technology and what is their motivation. There is so much to be optimistic about in artificial intelligence I would not be in this field if I did not believe that. But getting the best out of AI requires us to listen to all perspectives from all walks of life, engage with people from all zip codes across our country, embrace our global citizenship, and attract the best people from around the world.

I truly hope someday equal opportunity in AI will just be commonplace and not require such challenging discussions. It would be a lot more fun to make the case for why nonparametric belief propagation will become a better option than neural networks for more capable and explainable robot systems.

Chad Jenkins is an Associate Professor of Computer Science and Engineering and Associate Director of the Michigan Robotics Institute at the University of Michigan. He is a roboticist specializing in computer vision and human-robot interaction and leader of the Laboratory for Progress. He is a cofounder of BlackInComputing.org.

Originally posted here:

Before we put $100 billion into AI - VentureBeat

By now, we all know that more and more AI driven applications are introduced to the Radiology market, changing the diagnostic field and improving the diagnostic process. Not far behind, a substantial shift is happening in another critical medical segment the operating room.

More and more medical robotic companies are taking advantage of the new capabilities driven from AI and innovative computer vision algorithms, to provide solutions that add accuracy and stability, save time and improve supervision on many daily surgery procedures. Besides improving the surgery process, this AI revolution also shortens recovery time and reduces infection risks, by favoring minimally invasive methods, which expose only the minimal needed surface for the robotic arms operation.

"Our teams are developing state-of-the-art algorithmics solutions that make this revolution so real, giving surgical systems the capability to understand the medical scene, detect and monitor surgical tools during the procedure, offer a crystal clear real-time view of the treated organ, while allowing precise depth measurements and video supervision of the whole procedure indicating irregularities, missing tools or disposables and much more," explained Ron Soferman in a recent interview.

RSIP Vision, is a global leader in artificial intelligence (AI) and computer vision technology, is developing and providing these advanced modules to Medical Surgery Vendors allowing them to provide specific solutions for a variety of surgical operations, in the Orthopedics, Aesthetics and General Surgery fields.

RSIP Vision is headquartered in Jerusalem Israel, and has a U.S. office in San Jose, CA.

More information is available on the company website: http://www.rsipvision.com.

SOURCE RSIP Vision

http://www.rsipvision.com

See the rest here:

RSIP Vision CEO: AI in medical devices is reducing dependence on human skills and improving surgical procedures and outcome - PRNewswire

By

Published: 10 Aug 2020

If tech professionals expect their IoT or AI investments to continue to deliver value for many years to come, they must know how the futures of IoT and AI are inextricably intertwined.

The potential of AI for IoT is still developing. As more data accumulates from connected devices and gets input into AI algorithms, the effects of AI and IoT on human lives and jobs will grow. The combination of AI and IoT will lead to a highly automated future where a web of IoT technology and smart software entities work together to drive advancements and better inform decisions.

Right now, humans are part of the AI and IoT system, whether they input data into machine learning algorithms or give final approval of automated actions. But experts can envision a future where that won't be the case. The future of IoT will be a web of smart entities where analysis of real-time data from many endpoints feed into multiple models that empower smart devices to automatically act, according to Michael Wollowski, an associate professor of computer science and software engineering at Rose-Hulman Institute of Technology, and John McDonald, CEO of ClearObject.

Wollowski and McDonald authored Chapter 7 in the book Artificial Intelligence for the Internet of Everything -- written by several experts and edited by William Lawless, et al -- which explores potential applications of IoT technology and what intelligent automation will look like in the future. This chapter explores how the future of IoT cannot be seen without the context of other technological advancements, including predictive analytics. The book dives into different uses of AI for IoT, how multiple AI and IoT systems will collaborate to reach their full potential, and the challenges IoT and AI create for cybersecurity and understanding the complexity of the technology.

Below is an excerpt of the book: Chapter 7, "The web of smart entities -- aspects of a theory of the next generation of the internet of things," written by Wollowski and McDonald.

See original here:

Examine the future of AI for IoT to achieve business success - TechTarget

This article is part of ourreviews of AI research papers, a series of posts that explore the latest findings in artificial intelligence.

Human-level performance. Human-level accuracy. Those are terms you hear a lot from companies developing artificial intelligence systems, whether its facial recognition, object detection, or question answering. And to their credit, the recent years have seen many great products powered by AI algorithms, mostly thanks to advances in machine learning and deep learning.

But many of these comparisons only take into account the end-result of testing the deep learning algorithms on limited data sets. This approach can create false expectations about AI systems and yield dangerous results when they are entrusted with critical tasks.

In a recent study, a group of researchers from various German organizations and universities have highlighted the challenges of evaluating the performance of deep learning in processing visual data. In their paper, titled, The Notorious Difficulty of Comparing Human and Machine Perception, the researchers highlight the problems in current methods that compare deep neural networks and the human vision system.

In their research, the scientist conducted a series of experiments that dig beneath the surface of deep learning results and compare them to the workings of the human vision system. Their findings are reminder that we must be cautious when comparing AI to humans, even if it shows equal or better performance on the same task.

In the seemingly endless quest to reconstruct human perception, the field that has become known as computer vision, deep learning has so far yielded the most favorable results. Convolutional neural networks (CNN), an architecture often used in computer vision deep learning algorithms, are accomplishing tasks that were extremely difficult with traditional software.

However, comparing neural networks to the human perception remains a challenge. And this is partly because we still have a lot to learn about the human vision system and the human brain in general. The complex workings of deep learning systems also compound the problem. Deep neural networks work in very complicated ways that often confound their own creators.

In recent years, a body of research has tried to evaluate the inner workings of neural networks and their robustness in handling real-world situations. Despite a multitude of studies, comparing human and machine perception is not straightforward, the German researchers write in their paper.

In their study, the scientists focused on three areas to gauge how humans and deep neural networks process visual data.

The first test involves contour detection. In this experiment, both humans and AI participants must say whether an image contains a closed contour or not. The goal here is to understand whether deep learning algorithms can learn the concept of closed and open shapes, and whether they can detect them under various conditions.

For humans, a closed contour flanked by many open contours perceptually stands out. In contrast, detecting closed contours might be difficult for DNNs as they would presumably require a long-range contour integration, the researchers write.

For the experiment, the scientists used the ResNet-50, a popular convolutional neural network developed by AI researchers at Microsoft. They used transfer learning to finetune the AI model on 14,000 images of closed and open contours.

They then tested the AI on various examples that resembled the training data and gradually shifted in other directions. The initial findings showed that a well-trained neural network seems to grasp the idea of a closed contour. Even though the network was trained on a dataset that only contained shapes with straight lines, it could also performed well on curved lines.

These results suggest that our model did, in fact, learn the concept of open and closed contours and that it performs a similar contour integration-like process as humans, the scientists write.

However, further investigation showed that other changes that didnt affect human performance degraded the accuracy of the AI models results. For instance, changing the color and width of the lines caused a sudden drop in the accuracy of the deep learning model. The model also seemed to struggle with detecting shapes when they became larger than a certain size.

The neural network was also very sensitive to adversarial perturbations, carefully crafted changes that are imperceptible to the human eye but cause disruption in the behavior of machine learning systems.

To further investigate the decision-making process of the AI, the scientists used a Bag-of-Feature network, a technique that tries to localize the bits of data that contribute to the decision of a deep learning model. The analysis proved that there do exist local features such as an endpoint in conjunction with a short edge that can often give away the correct class label, the researchers found.

The second experiment tested the abilities of deep learning algorithms in abstract visual reasoning. The data used for the experiment is based on the Synthetic Visual Reasoning Test (SVRT), in which the AI must answer questions that require understanding of the relations between different shapes in the picture. The tests include same-different tasks (e.g., are two shapes in a picture identical?) and spatial tasks (e.g., is the smaller shape in the center of the larger shape?). A human observer would easily solve these problems.

For their experiment, the researchers use the ResNet-50 and tested how it performed with different sizes of training dataset. The results show that a pretrained model finetuned on 28,000 samples performs well both on same-different and spatial tasks. (Previous experiments trained a very small neural network on a million images.) The performance of the AI dropped as the researchers reduced the number of training examples, but degradation in same-different tasks was faster.

Same-different tasks require more training samples than spatial reasoning tasks, the researchers write, adding, this cannot be taken as evidence for systematic differences between feed-forward neural networks and the human visual system.

The researchers note that the human visual system is naturally pre-trained on large amounts of abstract visual reasoning tasks. This makes it unfair to test the deep learning model on a low-data regime, and it is almost impossible to draw solid conclusions about differences in the internal information processing of humans and AI.

It might very well be that the human visual system trained from scratch on the two types of tasks would exhibit a similar difference in sample efficiency as a ResNet-50, the researchers write.

The recognition gap is one of the most interesting tests of visual systems. Consider the following image. Can you tell what it is without scrolling further down?

Below is the zoomed-out view of the same image. Theres no question that its a cat. If I showed you a close-up of another part of the image (perhaps the ear), you might have had a greater chance of predicting what was in the image. We humans need to see a certain amount of overall shapes and patterns to be able to recognize an object in an image. The more you zoom in, the more features youre removing, and the harder it becomes to distinguish what is in the image.

Deep learning systems also operate on features, but they work in subtler ways. Neural networks sometimes the find minuscule features that are imperceptible to the human eye but remain detectable even when you zoom in very closely.

In their final experiment, the researchers tried to measure the recognition gap of deep neural networks by gradually zooming in images until the accuracy of the AI model started to degrade considerably.

Previous experiments show a large difference between the image recognition gap in humans and deep neural networks. But in their paper, the researchers point out that most previous tests on neural network recognition gaps are based on human-selected image patches. These patches favor the human vision system.

When they tested their deep learning models on machine-selected patches, the researchers obtained results that showed a similar gap in humans and AI.

These results highlight the importance of testing humans and machines on the exact same footing and of avoiding a human bias in the experiment design, the researchers write. All conditions, instructions and procedures should be as close as possible between humans and machines in order to ensure that all observed differences are due to inherently different decision strategies rather than differences in the testing procedure.

As our AI systems become more complex, we will have to develop more complex methods to test them. Previous work in the field shows that many of the popular benchmarks used to measure the accuracy of computer vision systems are misleading. The work by the German researchers is one of many efforts that attempt to measure artificial intelligence and better quantify the differences between AI and human intelligence. And they draw conclusions that can provide directions for future AI research.

The overarching challenge in comparison studies between humans and machines seems to be the strong internal human interpretation bias, the researchers write. Appropriate analysis tools and extensive cross checks such as variations in the network architecture, alignment of experimental procedures, generalization tests, adversarial examples and tests with constrained networks help rationalizing the interpretation of findings and put this internal bias into perspective. All in all, care has to be taken to not impose our human systematic bias when comparing human and machine perception.

Original post:

Computer vision: Why its hard to compare AI and human perception - TechTalks

A unique data-driven scientific approach to study and predict excessive drinking using social media and mobile-phone data has wonAndrew Schwartz, assistant professor in the Department of Computer Science, and his team a $2.5M award from the National Institutes of Health (NIH). Their research will develop an innovative approach utilizing data from texting and social media as well as mobile phone apps to better understand how unhealthy drinking manifests in daily life and push the envelope for the ability of Artificial Intelligence (AI) to predict human behaviors.

The cross-disciplinary team will build AI models that predict future drinking as well as future effects of drinking on an individuals mood among service industry workers. The award will be distributed over four years in collaboration withRichard Rosenthal, MD, Director of Addiction Psychiatry at Stony Brook Medicine, andChristine DeLorenzo, Associate Professor in the Departments of Biomedical Engineering and Psychiatry and a team at the University of Pennsylvania.

Andy is blazing the trail in advancing AI tools for tackling major health challenges, said Fotis Sotiropoulos, Dean,College of Engineering and Applied Sciences. His work is an ingenious approach using data-science tools, smart-phones and social media postings to identify early signs of alcohol abuse and alcoholism and guide interventions. This is AI-driven discovery and innovation at its very best!

With the aid of the teams methods, social media content can be collected and analyzed faster and cheaper, and present an unscripted, less biased psychological assessment. Traditional ecological studies of unhealthy drinking are done via costly and time-consuming phone interviews, which can also be subject to poor data quality and biases. Schwartz and his team will use their novel AI-based approach over social media text content. Samir Das, Chair of the Department of Computer Science, said: Andrew has very successfully applied large-scale data and text analysis techniques for important and timely human health and well-being applications with very impactful results.

We now know analysis of everyday language can cover a wide array of daily factors affecting individual health but their use over timespans is limited. The methods we will develop in this project should enable real-time study into how health plays out in the each individuals own words and bring about the possibility for personalized mental health care, said Schwartz.

The technology will be developed with a focus on a population of frontline restaurant workers bartenders and servers a group that has among the highest rates of heavy alcohol use of all professions. This unhealthy drinking (defined as seven drinks a week for women and 14 for men, according to the National Institute on Alcohol Abuse and Alcoholism) creates the potential for extensive negative consequences related to work performance, relationships, and physical and psychological health. For example, the team will look at the effect of empathy, as measured through language. Psychologists on the team note that empathy can be both health-promoting (beneficial) and health-threatening (depleting). Distinguishing beneficial versus depleting empathy is an example where AI can capture something difficult to get at through questions. Its also a dimension of human psychology suspected to play a role in stress on servers and bartenders since they often listen to customers problems and provide advice, which could have a negative effect on them.

The teams research will include development of:

Schwartzs research has also focused on the impact of social media to predict mental and physical health issues. He is also using Twitter to studyCOVID-19and before that he focused ondepression and social media posts.

About the ResearcherAndrew Schwartzis an Assistant Professor in the Department of Computer Science and a faculty member of theInstitute for AI-Driven Discovery and Innovation. His research utilizes natural language processing and machine learning techniques to focus on large-scale language analysis for health and social sciences.

Dan Olawski

See the original post here:

$2.5 Million NIH Grant Will Support AI Approach to Study and Predict Excessive Drinking | | SBU News - Stony Brook News

Art imitating art? That isnt quite the case here, but a new algorithm developed by the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) in conjunction with Microsoft searchers for art that looks like other artworks. Called MosAIc, the algorithm uses artificial intelligence to search for works that are similar in nature or similar in concept, from various cultures and time periods, to a specified artwork.

Led by Mark Hamilton, a CSAIL PhD student, MosAIc was inspired by Rembrandt and Velzquez, an exhibition centred on Rembrandt and Diego Velzquez shown at Amsterdams Rijksmuseum that wrapped up in January of this year. The exhibition showcased the similarities between works by the 17th century masters, one Dutch and one Spanish, among others, although there is no evidence to support them the artists would have been aware of one another. One such comparison included in the exhibition was of Francisco de Zurbarns The Martyrdom of Saint Serapionand Jan AsselijnsThe Threatened Swan. The pair of paintings piqued Hamiltons interest, and soon, MosAIc, which brings together unlikely, yet uncanny artworks, was born.

AI has become more ingrained in the art world in recent years, as artworks created by AI have started to take off, even if for some theyre simply fascinating oddities. Instead, MosAIc works as a Conditional Image Retrieval (CIR) application that needs just one image to get started, which sets it apart from other existing projects, like X Degrees of Separation by Google, that seek similar goals.

MosAIc was trained using New York Citys Metropolitan Museum of Art and the Rijksmuseums open-access collections. Using a treelike data-structure, called a conditional KNN tree, MosAIc searches its stores for artworks that are related to the original image. The related works are categorised into two branches, either media or culture, which then allows users to adjust the parameters in order to find objects from a specific culture or medium of art to narrow down, or prune, their search. MosAIc spans centuries of works, creating a network of art from various cultures and artists who were working a plethora of mediums. Additionally, Hamilton and his team worked to allow MosAIc to link artworks that are were similar in meaning and theme, not only colour and style.

Every time I use the algorithm, Hamilton told ArtNet News, I find surprises.

Hamilton, along with Stephanie Fu, William T. Freeman, and Mindren Lu, all of MIT, have published a paper on their research into CIR, such as MosAIc, discussing its aims and abilities. The team doesnt expect MosAIc to replace curators, recognising the invaluable expertise they and art historians bring in discovering deeper meaning and context amongst artists and their works; however, they hope MosAIc will become a resource for experts planning exhibitions and perhaps even those in other areas.

Going forward, we hope this work inspires others to think about how tools from information retrieval can help other fields like the arts, humanities, social science, and medicine, Hamilton said in MIT News. These fields are rich with information that has never been processed with these techniques and can be a source for great inspiration for both computer scientists and domain experts. This work can be expanded in terms of new datasets, new types of queries, and new ways to understand the connections between works.

MosAIc also surprised the team in its ability to pick out the blind spots of GANs, or generative adversarial networks that are used in making so called deepfakes. While deepfakes can be an informational and engaging tool, think of the way AI brought the Mona Lisa or Salvador Dal, himself, to life they have also been linked to the creation of fake news and false propaganda. In being able to flag such GANs, MosAIc could be beneficial in identifying AI-generated fakes that have become more prevalent in the last year or so and will most likely only become more a part of day-to-day life in the future.

Related

The rest is here:

MIT develops AI that identifies similarities between unrelated artworks, spanning centuries, artists, and mediums - Art Critique