Archive For The “Artificial Intelligence” Category
Israeli startup Aidoc, a developer of artificial intelligence (AI)-powered software that analyzes medical images, announced on Wednesday that it received Food and Drug Administration (FDA) clearance for a solution that flags cases of Pulmonary Embolism (PE) in chest scans for radiologists.
Portions of this article were originally reported in NoCamels.com
Aidoc has CE (Conformité Européenne) marking for the identification and triage of pulmonary embolism (PE) in CT pulmonary angiograms, and FDA approval to scan images for brain hemorrhages.
The latest approval came a month after Aidoc secured $27 million in a Series B round led by Square Peg Capital. Founded in 2016 by Guy Reiner, Elad Walach, and Michael Braginsky, the company has raised some $40 million to date.
Aidoc’s technology assists radiologists in expediting problem-spot detection through specific parameters such as neuron-concentration, fluid-flow, and bone-density in the brain, spine, abdomen, and chest. Aidoc says its solutions cut the time from scan to diagnosis for some patients from hours to under five minutes, speeding up treatment and improving prognosis.
“What really excites us about this clearance is that it paves the way towards scalable product expansion,” Walach, who serves as Aidoc CEO, said in a statement. “We strive to provide our customers with comprehensive end-to-end solutions and have put a lot of effort in developing a scalable AI platform.”
Walach said the company has eight more solutions in active clinical trials.
Apparently, Facebook is of the opinion that what it needs is more artificial intelligence (AI). The social media giant just announced the formation of the Data.AI group, a Tel Aviv-based AI team that will focus on machine learning, and developing tools to assist with deeper analytical insights.
Featured story. Artificial Intelligence, part IV.
According to a recent report, AI is a major source of growth for the Israeli tech industry.
Israel is home to over 1,000 companies, academic research centers, and multinational R&D centers specializing in AI, including those that develop core AI technologies, as well as those that utilize AI technologies for their vertical-related products such as in healthcare, cybersecurity, automotive, and manufacturing among others, according to the Start-Up Nation Central report.
Israeli companies specializing in artificial intelligence raised nearly 40 percent of the total venture capital funds raised by the Israeli tech ecosystem for 2018, despite accounting for just 17 percent of the total number of technology companies in the country, the report noted.
The SNC report noted that a number of events in 2018 boosted the AI ecosystem in Israel, including the launch of a new Center for Artificial Intelligence by Intel and the Technion-Israel Institute of Technology, and the announcement by US tech giant Nvidia (which acquired Israel’s Mellanox Technologies last month for $6.9 billion) that it too was opening a new AI research center.
A number of high-profile AI products developed by Israeli teams working for multinationals were also unveiled this year. In May, Google came out with Google Duplex, a system for conducting natural sounding conversations developed by Yaniv Leviathan, principal engineer, and Yossi Matias, vice president of engineering and the managing director of Google’s R&D Center in Tel Aviv. And in July 2018, IBM unveiled Project Debater, a system powered by artificial intelligence (AI) that can debate humans, developed over six years in IBM’s Haifa research division in Israel.
Earlier this year, the Israel Innovation Authority (IIA) warned that despite industry achievements, Israel was lagging behind other countries regarding investment in AI infrastructures and urgently needed a national AI strategy to keep its edge. The IIA called for the consolidation of all sectors – government, academia, and industry – to establish a vision and a strategy on AI for the Israeli economy.
A new technological breakthrough is using AI and facial analysis to make it easier to diagnose genetic disorders. DeepGestalt is a deep learning technology created by a team of Israeli and American researchers and computer scientists for the FDNA company based in Boston. The company specializes in building AI-based, next-generation phenotyping (NGP) technologies to “capture, structure and analyze complex human physiological data to produce actionable genomic insights.”
Portions of this article were originally reported in NoCamels.com
DeepGestalt uses novel facial analysis to study photographs of faces and help doctors narrow down the possibilities. While some genetic disorders are easy to diagnose based on facial features, with over 7,000 distinct rare diseases affecting some 350 million people globally, according to the World Health Organization, it can also take years – and dozens of doctor’s appointments – to identify a syndrome.
“With today’s workflow, it can mean about six years for a diagnosis. If you have data in the first year, you can improve a child’s life tremendously. It is very frustrating for a family not to know the diagnosis,” Yaron Gurovich, Chief Technology Officer at FDNA and an Israeli expert in computer vision, tells NoCamels. “Even if you don’t have a cure, to know what to expect, to know what you’re dealing with helps you manage tomorrow.”
DeepGestalt — a combination of the words ‘deep’ for deep learning and the German word ‘gestalt’ which is a pattern of physical phenomena — is a novel facial analysis framework that highlights the facial phenotypes of hundreds of diseases and genetic variations.
According to the Rare Disease Day organization, 1 in 20 people will live with a rare disease at some point in their life. And while this number is high, there is no cure for the majority of rare diseases and many go undiagnosed.
“For years, we’ve relied solely on the ability of medical professionals to identify genetically linked disease. We’ve finally reached a reality where this work can be augmented by AI, and we’re on track to continue developing leading AI frameworks using clinical notes, medical images, and video and voice recordings to further enhance phenotyping in the years to come,” Dekel Gelbman, CEO of FDNA, said in a statement.
DeepGestalt’s neural network is trained on a dataset of over 150,000 patients, curated through Face2Gene, a community-driven phenotyping platform. The researchers trained DeepGestalt on 17,000 images and watched as it correctly labeled more than 200 genetic syndromes.
In another test, the artificial intelligence technology sifted through another 502 photographs to identify potential genetic disorders.
DeepGestalt provided the correct answer 91 percent of the time.
Indeed, FDNA, a leader in artificial intelligence and precision medicine, in collaboration with a team of scientists and researchers, published a milestone study earlier this year, entitled “Identifying Facial Phenotypes of Genetic Disorders Using Deep Learning” in the peer-reviewed journal Nature Medicine.
Pharmaceutical R&D has huge barriers to entry. The cost is just too expensive for ambitious startups to even consider, on average about $2.5 billion just to bring a new drug to market. And there’s also a public interest cost which is seldom discussed. Namely, the cynical fact that drugs that can’t make much money are never developed. Hopefully, both of these impediments will lessen in the near future. Once again, artificial intelligence (AI) and machine learning (ML) are at the forefront in the drive reduce the time and cost to develop new drugs. How? By letting the molecules do the talking.
The following article was originally published in NoCamels.com
The cost of developing a new pharmaceutical drug, from the research and development stage to market approval, runs at about $2.6 billion, according to a 2014 report published by the Tufts Center for the Study of Drug Development (CSDD) cited by the Scientific American. It also takes between 10 to 15 years.
Israeli scientists say they have developed a revolutionary smart method to discover and develop new drugs, based on artificial intelligence and machine learning, that will dramatically shorten preparation time and reduce costs.Dr. Kira Radinsky, a renown data scientist and a visiting professor at the Technion – Israel Institute of Technology, and Shahar Harel, a PhD student at the university’s computer science department, presented their system late last month at the KDD 2018 conference in London, an annual event on Big Data and Machine Learning that draws prominent world academics and industry leaders.
Radinsky and Harel’s system seeks to tap into the modern-day, computerized processes of screening and selecting molecules with the greatest therapeutic potential – of which there are more than stars in the galaxy, making this an enormous task.
Their working hypothesis is that drug development “vocabulary” is similar to that of a natural language.
Harel said in a university statement that the system he and Radinsky developed, founded on artificial intelligence (AI) and deep learning, “acquired this language based on hundreds of thousands of molecules.”
“We are essentially presenting here an algorithm which addresses the creative stage of drug development – the molecule discovery stage,” said Harel. “This capacity leans upon our mathematical innovation, which enables the computer to understand the chemical language and to generate new molecules based upon a prototype.”
The researchers instructed the system to propose 1000 drugs based upon old drugs and were surprised to discover that 35 of the new drugs generated by the system are existing, FDA-approved drugs developed and approved after 1950. Radinsky said in a statement that the system “is not only a means of streamlining existing methods but also entirely new drug development and scientific practice paradigms.”
“Instead of seeking out specific correlations based upon hypotheses we formulate, we allow the computer to identify these connections from within a massive sample size, without guidance. The computer is not smarter than man, but it can cope with huge amounts of data and find unexpected correlations,” she added.
Radinsky indicated that a similar computerized process is how, in another study, the scientists managed to find the unknown side effects of various drugs and drug combinations.
“This is a novel type of science which is not built upon hypotheses tested in an experiment, rather, upon data that generated the research hypothesis,” she said.
The Technion said in a statement that the breakthrough is particularly significant in light of Eroom’s Law, which asserts that the number of new drugs approved by the FDA should decline at a rate of approximately 50 percent every nine years. The term was coined in 2012 in an article published in Nature Reviews Drug Discovery and is a reverse order of Moore, the name of Gordon Moore, one of the founders of Intel. Moore observed that the number of transistors in a dense integrated circuit doubles every two years. In contrast, Eroom’s Law notes that each year, fewer and fewer drugs are marketed.
Dr. Radinksy projects that “this new development will accelerate and reduce costs of development of new and effective drugs, thereby shortening the time patients will have to wait for the drugs. In addition, this breakthrough is expected to lead to the development of drugs that would not have been generated with the conventional pharmacological paradigm.”
The system is currently being deployed for use in collaboration with pharmaceutical companies to further analyze the additional generated molecules, the scientists said.
I have written extensively about artificial intelligence (AI), noting its far-reaching tentacles, diverse applications, and ubiquities. But there’s a companion platform that has been raging for a few years now, and that platform is Blockchain.
Unless you’re a tech geek, you probably have a cursory at best understanding, having heard of it in news reports. The best way to put Blockchain into context is to understand its most popular application: Bitcoin. This cryptocurrency has caused volatile swings in the financial markets, has caused the biggest banks and lending institutions to take notice, even throwing millions in R&D to bone up on the technology.
More on Diane Israel.
What’s important to understand is the relationship between Bitcoin (and other lesser known cryptocurrencies) and Blockchain. Bitcoin runs on blockchain and currently needs blockchain to function.
As such, the marriage between AI and Blockchain is a natural one, although not necessary. The appeal, however, is the security features that blockchain promises, and the egalitarian way in which information is stored and distributed.
The Israeli Connection
Israeli startups are at the forefront of both of these sectors, receiving notable attention and investments from global players which are further propelling Israeli development in these industries.
As recently reported in NoCamels.com:
In Israel, the average investment per deal in AI grew five times in value, from $2 million in 2016 to $10.2 million in 2017. Subsequently, the growth in this sector is reflected in the overall investment numbers for AI in Israel, with the market growing from $55 million in 2016 to $472 million in 2017, according to the Geektime Annual Report 2017: Startups and venture capital in Israel, published in January.
A major setback.
For a few years now, financial institutions have been experimenting with Blockchain for its “very strong” security features. Not so fast, however. What was lost by most was FPI special investigator Robert Mueller’s recent charges filed against senior Russian military leaders, many of whom take their orders directly from Vladimir Putin. What the filing showed, in great detail, is how Mueller’s team was able to deconstruct the way in which the Russians used Bitcoin to pay for everything from website hosting, domain registration, and even paying for Facebook ads. According to many Bitcoin experts, the advantage of Bitcoin/Blockchain is its ability to remain anonymous throughout the transaction process. And while we still don’t know how Mueller’s team was able to attribute a Bitcoin account directly to its true owner is proof positive that the technology is vulnerable.
Netting it out.
What this means is simple. Blockchain may have some appealing features for the marketplace, but security isn’t one of them. As such, the fledgling marriage between AI and Blockchain may be short-lived and was never a necessary one.
While at times the evolution of technological innovation may seem chaotic with no clear purpose, goal or objective — many new technologies seem to come out of nowhere — there is an unseen hand at play. Adam Smith’s Wealth of Nations foretells of this somewhat mystical phenomenon whereby markets have their own agency, filling in market gaps as if a transcendental being was overseeing our economy. What Smith and many others who followed him neglected to notice, whether intentionally or not, is that real people with keen awareness of present conditions coupled with future need are these mysterious beings. In other words, we have discovered this unseen hand, and it’s us!
I’ll use applications that just about everyone is aware of to demonstrate who all of this works. It’s a bit of an oversimplification but appropriate for this demonstration.
- Microsoft Word released as a standalone application
- Microsoft Excel released as a standalone application
- Microsoft PowerPoint released as a standalone application
Then Microsoft Office released comprising all three with true integration that made interfacing with all three rather easy. In other words, innovations begin as separate entities but eventually and naturally consolidate into one integrated application.
I mention this because we are witnessing the same sort of stovepipe development and consolidation happening right now. However, this phenomenon is no longer constrained to applications but rather more vague concepts such as content and speed.
On the content side, artificial intelligence (AI) is the driving force. Note that there is real category confusion about AI that is to be expected. Legacy labels such as neural networks and machine learning are becoming meaningless because each overlap and can rightly be called AI, which depicts a consolidation of applications in real-time. Now couple that with what AI needs to allow for greater capabilities, ones that science fiction describes, and there we have it. Speed. And this increased speed, actually 20to 50 times faster than its predecessor, is coming through 5g wireless.
Long story short, if you want to see the future of tech innovation, keep your eyes on AI and 5G. Throw in the peripheral technologies of the Internet of Things (IoT), and the picture becomes clear.
IBM may have missed the mark on becoming a PC giant after its poorly calculated entry into the market a few decades ago. The computer-before-there-was-a-computer market giant also was a bust with ill-fated entries into PC operating systems, OS2 RIP! And quite frankly more of the same from its entrants into Web servers, e-commerce, content management…, it’s a very long list of failures.
But for artificial intelligence (AI), IBM has found its niche, which itself is ironic since a computer giant is not supposed to be a niche player. From the early days of Deep Blue, the first computer to beat reigning world chess champion Gary Kasparov back in 1996.
Since then IBM’s AI has gotten a lot smarter, and with the help of its Israeli IBM Haifa division, it’s debating humans in situations for which the rules are not nearly as structured as Chess as the following article excerpt from NoCamels.com explains.
Dubbed PROJECT DEBATER, it was developed over six years in IBM’s Haifa research division in Israel.
At the unveiling two weeks ago in San Francisco, the system engaged in its first-ever live, public debate. Its opponents were two Israeli debate champions. Israel’s 2016 debate champion Noa Ovadia took on the system for a discussion on whether space exploration should be subsidized by the government. Dan Zafrir, a professional debater, argued Project Debater on the value of telemedicine and whether it should be used more widely.
Each side delivered a four-minute opening statement, a four-minute rebuttal, and a two-minute summary, according to a June 18 post by IBM Research Director Arvind Krishna
The humans were said to have won, but by a close call. According to an audience survey cited by Krishna in an interview with Fox News, the computer lacked the persuasive speaking nuances of the debate champs but possessed more knowledge on the topics. Krishna wrote that IBM “selected from a curated list of topics to ensure a meaningful debate. But Project Debater was never trained on the topics.”
This week, Project Debater performed once again against two human debaters, this time in Israel where the team behind the project proudly displayed it.
At the event at IBM’s Givatayim offices held for local press, the system this time challenged Israeli professional debaters Yaar Bach and Hayah Goldlist-Eichler on mass surveillance methods, and genetic engineering, respectively.
IBM’s Israel CEO and country manager Daniel Melka told the audience that the company developed “very special technology” that is “a significant milestone in the development of Artificial Intelligence technology,” according to the Times of Israel.
In a video presentation ahead of the unveiling, Noam Slonim, the principal investigator of Project Debater and senior technical staff member (STSM) at the IBM Haifa Research Lab, said the goal of the project was “to demonstrate that we can have a meaningful and viable discussion between men and machine.”
Project Debater, Krishna wrote, “moves us a big step closer to one of the great boundaries in AI: mastering language. It is the latest in a long line of major AI innovations at IBM, which also include “Deep Blue,” the IBM system that took on chess world champion Garry Kasparov in 1997, and IBM Watson, which beat the top human champions on Jeopardy! in 2011.”
IBM’s recent developments in machine thinking surpass that of existing products such as Apple’s Siri and Amazon’s Alexa. These devices primarily recite rote information, whereas Project Debater uses facts to reason and construct arguments on topics that have no right or wrong answers. According to IBM, the technology accomplishes this through first recognizing opponents’ arguments through Watson Speech to Text. Then, it identifies relevant expressions in its database of hundreds of millions of articles from trusted journals and magazines. Lastly, it eliminates redundancies, prioritizes arguments and composes coherent English speech.
“Subsidizing space exploration is like investing in really good tires,” Project Debater rebutted Ovadia in the government-sponsored space research debate in San Francisco. “It may not be fun to spend the extra money, but ultimately you know both you and everyone else on the road will be better off.” It further argued that space exploration also inspires the younger generation to pursue careers in science and technology.
The computer also attempted to make jokes during the debate. “You are speaking at the extremely fast rate of 218 words per minute. There is no need to hurry,” Project Debater told Ovadia.
Up against Zafrir in the telemedicine debate, the system admonished its opponent saying: “Fighting technology means fighting human ingenuity.” And in the debate this week against Goldlist-Eichler, who, for the sake of argument expressed her suspicions of the safety of technological advancement, Project Debater said: “I can’t say this is getting on my nerves, because I don’t have any.”
The project is being hailed as the onset of a new era for human-machine interaction. Krishna says IBM’s mission was to develop broad AI that learns across different disciplines to augment human intelligence.
And Krishna said Project Debater could become “the ultimate fact-based sounding board without the bias that often comes from humans.”
Project Debater has its limitations. The system is currently programmed to follow a strict 20-minute debate format for 100 topics, according to The New York Times.
Furthermore, Wired magazine reported that Project Debater sometimes failed to address certain points and to construct rebuttals in line with the opponents, and provide real-life context for its arguments.
Krishna acknowledged that building the system was a “remarkably difficult and complex challenge,” and that it makes mistakes, “just like people.”
Though the Israeli team built Project Debater with three ground-breaking AI capabilities (data-driven speech writing and delivery, listening comprehension that can identify key claims hidden within long continuous spoken language, and modeling human dilemmas in a unique knowledge graph to enable principled arguments), the system must still learn to “adapt to human rationale and propose lines of argument that people can follow.”
“Debate rules stem from a human culture of discussion and are not arbitrary, and the value of arguments is often inherently subjective…In debate, AI must learn to navigate our messy, unstructured human world as it is – not by using a pre-defined set of rules, as in a board game,” he wrote.
While PROJECT DEBATER technology lost the debate, it demonstrated more knowledge than its two human counterparts. So what caused PD to lose? Influence and persuasion. PD simply lacked the subtleties of language and nuanced delivery to maximize its influence. In other words, it lost on style points, which tells us a lot. Indeed, style does matter when it comes to persuading others to accept the knowledge being presented.
In this segment on artificial intelligence (AI), part four, we’ll look at the augmented human being, part human, part machine. And don’t laugh. Computer or robotic-assisted devices are being used to augment the human condition right now. For but one example, see my previous story on exoskeleton technology.
Another slick piece of wearables allows legally blind people to read newspaper and magazines, or product labels in a grocery store, even the money they take out of their pocket to pay the cashier, using artificial visualization technology.
As neuroscientists unleash the mysteries and power of the human brain while, at the same time, AI researchers build programs that get smart and smarter, even to the point where they become autonomous learners, human anatomy and robotics, along with AI software, will converge into human/machine hybrids, some of which will have more human characteristics than others. In other words, if we live long enough, say twenty more years, we may actually meet Mr. Spock, or a reasonable facsimile thereof.
Some of my academic friends who are working on this exciting future are not as enthusiastic as you would think. Many fear that the ethics will not keep pace with the technology, that we will create, arguably, a new species whose rights and freedoms will not comport with our justice system as it is today. Others are concerned about the economic value of people in an age where machines and computers will do almost all of the work. What are we going to do with 5 billion in surplus labor for which there will never be a job? Without income potential yet still constantly need to consume goods and services, how will the contribute to the betterment of our species and our world? There are no good answers for any of this yet. But there certainly are many grave concerns over them and many others.
But with all the ethical, economic and social concerns over AI, what most scientists are most anxious over is the notion of singularity. Singularity, as it pertains to AI, is the moment in the future whereby computers will become not only smarter than humans (and their programmers) but autonomous as well. If you haven’t guessed by now, they’re talking about the master/slave relationship between man and machine flipping. How this would exactly happen, nobody really knows. The anxiety of such a time is difficult to imagine. But it’s almost definitely only a few decades away. And while I may be naive, if a bunch of Mr. Spocks started running our world, it’s hard to imagine how that wouldn’t be an improvement.
In this segment, we’ll take a look at the practical applications of artificial intelligence (AI) today and what is right around the corner.
Recall the days on the evening news when reporters would interview a trader on the floor of the New York Stock Exchange to cover the impact of a trade war, a real war, or more the more disappointing earnings report from a blue chip company? Remember all those white middle-aged men in the middle of the pit gyrating most of every imaginable histrionic while shouting in desperation in secret Wall Street code and hand signals resembling gang signs with perspiration spraying everywhere?
You may not have realized it but those days are gone. Wall Street today is basically a working museum. The trading pits are gone. The only piece of the good old days that remains is the opening and closing bells which still does air on TV on occasion. The pit trader’s job has been lost to automation. Computer algorithms now do that job, and more efficiently too. But now that all (or most) human emotion has been jettisoned from the trade when unexpected events happen it’s all up to the computers to determine the best moves to make, all of which have been decided beforehand.
The same thing is true when you order up an Uber. The nearest pool of drivers is automatically petitioned. And the driver the gets the bid simply is verified of the task via a computer algorithm. Pickup and destination logistics, as well as price, are all determined by computers. As I write this, Uber has a new what I’ll call the “drunk algorithm” to determine the likelihood of a customer of having a few too many when ordering an Uber. The algorithm looks for, among other things, common typing mistakes, language used, location, and who knows what else, to determine their mental state and recommend an Uber driver more experience with unsober customers.
Advanced online marketing techniques use data analytics and other Big data to find predictive correlations between a consumers marital status and their likelihood to drink beer, and on what day, and what kind of beer. Men living in Nevada who have been recently divorced (say for less than two years) are more likely to buy a six-pack of beer on a Thursday, for instance. Knowing this, a Miller Lite advertisement will appear between 2 and 3 pm when they hit CNN.com to check the latest news.
These are but just a few examples of where AI is today. Imagine where it will be in five years! The only thing that can stop it is consumer insistence upon greater controls over their privacy concerns. But don’t hold your breath on that one. Most of us have already decided, albeit unwittingly, that the conveniences of the digital age outweigh the costs of giving up a bit of our privacy. In other words, we have traded away some of our privacy for its exchange value. And this is something we do a lot more than we would like to admit. You may recoil to the idea of having a microchip inserted into your person right now, but in five years you may find yourself opting into such a voluntary problem. Why? Because you may no longer need to remember your wallet, your keys, your passport, credit cards, rewards cards, pin numbers, or passwords. Pretty convenient, huh?
In this second part of the implications of artificial intelligence segment, we take a look at what is intelligence and can computers think, reason, and learn.
It was 1996. IBM’s “Deep Blue” supercomputer was to do battle with reigning world chess champion, Gary Kasparov, using standard rules of chess. Spoiler alert. Deep Blue won its first game against Kasparov on February 10, 1996, when it defeated him in the first game of a six-game match. Kasparov, however, rallied and won the next three, then drew two of the five games, ultimately defeating Big Blue by a score of 4 to 2. A year later, Big Blue was upgraded and this time defeating the reigning world champion narrowly.
Bear in mind, this was over twenty years ago. Computer processing capabilities were in the stone ages compared to where they are today, and today’s computing will follow the same fate with the future.
Was the computer thinking? Well if we consider thought to be the ability to reason correctly for the purpose of achieving some future goal, then yes. But was this reasoning ability trained by humans to reason well? Sure enough. But isn’t that the same thing our parents and teachers do when we are in our formative years (and beyond)? Perhaps the biggest difference is that humans teach other humans a mix of rational and irrational thought process whereas computers are thought “pure reason” much in the spirit of Mr. Spock in Star Trek.
For me what’s interesting is whether teaching a computer or teaching a human is fundamentally different. And whether the hardware and software used to think, reason, and learn is fundamentally different in how it processes information. What neuroscientists and computer scientists tell me — and I must defer to them for I am neither — is that computers were created by people and therefore the human way of reasoning was built into each of them. Whether this was conscious or not is a matter of debate and an important one. What both have in common is that they both learn through induction and inference. The chief difference is how the emotional content and influence on human behavior can cloud our ability to learn and reason. Computers have no such capability. Just imagine if they did! There would likely be no reason for this conversation if they could. And would we really want to rely on them to run our power grid, airline traffic, or our satellite communications? Then again, if all satellite communications crashed for a few hours it all could be attributable to the computers having a “bad day”.
I think we’re better off with computers has purely intellectual devices.