Is AI Riding a One-Trick Pony? – MIT Technology Review

Is AI Riding a One-Trick Pony? – MIT Technology Review

I’m standing in what is soon to be the center of the world, or is perhaps just a very large room on the seventh floor of a gleaming tower in downtown Toronto. Showing me around is Jordan Jacobs, who cofounded this place: the nascent Vector Institute, which opens its doors this fall and which is aiming to become the global epicenter of artificial intelligence. Indeed, backprop wasn’t discovered by probing deep into the brain, decoding thought itself; it grew out of models of how animals learn by trial and error in old classical-conditioning experiments. And most of the big leaps that came about as it developed didn’t involve some new insight about neuroscience; they were technical improvements, reached by years of mathematics and engineering. David Duvenaud, an assistant professor in the same department as Hinton at the University of Toronto, says deep learning has been somewhat like engineering before physics. “Someone writes a paper and says, ‘I made this bridge and it stood up!’ Another guy has a paper: ‘I made this bridge and it fell down—but then I added pillars, and then it stayed up.’ Then pillars are a hot new thing. Someone comes up with arches, and it’s like, ‘Arches are great!’” With physics, he says, “you can actually understand what’s going to work and why.” Only recently, he says, have we begun to move into that phase of actual understanding with artificial intelligence. And though we’ve started to get a better handle on what kinds of changes will improve deep-learning systems, we’re still largely in the dark about how those systems work, or whether they could ever add up to something as powerful as the human mind. Recommended for You If you want to see the next big thing, something that could form the basis of machines with a much more flexible intelligence, you should probably check out research that resembles what you would’ve found had you encountered backprop in the ’80s: smart people plugging away on ideas that don’t really work yet. A few months ago I went to the Center for Minds, Brains, and Machines, a multi-institutional effort headquartered at MIT, to watch a friend of mine, Eyal Dechter, defend his dissertation in cognitive science. On the screen was a picture of Ruby, and next to it one of Susannah as a baby. On the way out of the room, she wheeled a toy stroller behind her mom and yelled “Good luck, Daddy!” over her shoulder. “Vámanos!” she said finally. “The fact that it doesn’t work is just a temporary annoyance.” Eyal started his talk with a beguiling question: How is it that Susannah, after two years of experience, can learn to talk, to play, to follow stories? Backprop, in the words of Jon Cohen, a computational psychologist at Princeton, is “what all of deep learning is based on—literally everything.” When you boil it down, AI today is deep learning, and deep learning is backprop—which is amazing, considering that backprop is more than 30 years old. But it’s not as if you learn to make a soufflé by learning every one of the program’s zillion micro-instructions, like “Rotate your elbow 30 degrees, then look down at the countertop, then extend your pointer finger, then …” If you had to do that for every new task, learning would be too hard, and you’d be stuck with what you already know. Instead, we cast the program in terms of high-level steps, like “Whip the egg whites,” which are themselves composed of subprograms, like “Crack the eggs” and “Separate out the yolks.” Computers don’t do this, and that is a big part of the reason they’re dumb. Unlike a computer, she’ll have a model in her mind about how the whole world works. “It’s sort of incredible to me that people are scared of computers taking jobs,” Eyal says. “It’s not that computers can’t replace lawyers because lawyers do really complicated things. We’re so far.” A real intelligence doesn’t break when you slightly change the requirements of the problem it’s trying to solve. And the key part of Eyal’s thesis was his demonstration, in principle, of how you might get a computer to work that way: to fluidly apply what it already knows to new tasks, to quickly bootstrap its way from knowing almost nothing about a new domain to being an expert. It’s worth understanding how that happened—how a technique could lie in wait for so long and then cause such an explosion—because once you understand the story of backprop, you’ll start to understand the current moment in AI, and in particular the fact that maybe we’re not actually at the beginning of a revolution. It gets a computer to function somewhat like a programmer who builds up a library of reusable, modular components on the way to building more and more complex programs. Without being told anything about a new domain, the computer tries to structure knowledge about it just by playing around, consolidating what it’s found, and playing around some more, the way a human child does. His advisor, Joshua Tenenbaum, is one of the most highly cited researchers in AI. Tenenbaum’s name came up in half the conversations I had with other scientists. Some of the key people at DeepMind—the team behind AlphaGo, which shocked computer scientists by beating a world champion player in the complex game of Go in 2016—had worked as his postdocs. He’s involved with a startup that’s trying to give self-driving cars some intuition about basic physics and other drivers’ intentions, so they can better anticipate what would happen in a situation they’ve never seen before, like when a truck jackknifes in front of them or when someone tries to merge very aggressively. Eyal’s thesis doesn’t yet translate into those kinds of practical applications, let alone any programs that would make headlines for besting a human. “The problems Eyal’s working on “are just really, really hard,” Tenenbaum said. “It’s gonna take many, many generations.” Tenenbaum has long, curly, whitening hair, and when we sat down for coffee he had on a button-down shirt with black slacks. He hopes the same thing might happen with his own work and that of his students, “but it might take another couple decades.” As for Hinton, he is convinced that overcoming AI’s limitations involves building “a bridge between computer science and biology.” Backprop was, in this view, a triumph of biologically inspired computation; the idea initially came not from engineering but from psychology. But this was the same situation he’d been in with backprop for nearly 30 years. “This thing just has to be right,” he says about the capsule theory, laughing at his own boldness. “And the fact that it doesn’t work is just a temporary annoyance.” James Somers is a writer and programmer based in New York City. Vindication The walk from the Vector Institute to Hinton’s office at Google, where he spends most of his time (he is now an emeritus professor at the University of Toronto), is a kind of living advertisement for the city, at least in the summertime. You can understand why Hinton, who is originally from the U.K., moved here in the 1980s after working at Carnegie Mellon University in Pittsburgh. We’re in Toronto because Geoffrey Hinton is in Toronto, and Geoffrey Hinton is the father of “deep learning,” the technique behind the current excitement about AI. “In 30 years we’re going to look back and say Geoff is Einstein—of AI, deep learning, the thing that we’re calling AI,” Jacobs says. Toronto was built on top of forested ravines, and it’s said to be “a city within a park”; as it’s been urbanized, the local government has set strict restrictions to maintain the tree canopy. Toronto is the fourth-largest city in North America (after Mexico City, New York, and L.A.), and its most diverse: more than half the population was born outside Canada. There’s free health care and good public schools, the people are friendly, and the political order is relatively left-­leaning and stable; and this stuff draws people like Hinton, who says he left the U.S. because of the Iran-Contra affair. It’s one of the first things we talk about when I go to meet him, just before lunch. “Most people at CMU thought it was perfectly reasonable for the U.S. to invade Nicaragua,” he says. “They somehow thought they owned it.” He tells me that he had a big breakthrough recently on a project: “getting a very good junior engineer who’s working with me,” a woman named Sara Sabour. Google’s Toronto office scooped her up. Hinton, who is 69 years old, has the kind, lean, English-looking face of the Big Friendly Giant, with a thin mouth, big ears, and a proud nose. Of the AI researchers at the top of the field, Hinton has more citations than the next three combined. He was born in Wimbledon, England, and sounds, when he talks, like the narrator of a children’s book about science: curious, engaging, eager to explain things. He stands the whole time we talk, because, as it turns out, sitting is too painful. “I sat down in June of 2005 and it was a mistake,” he tells me, letting the bizarre line land before explaining that a disc in his back gives him trouble. It means he can’t fly, and earlier that day he’d had to bring a contraption that looked like a surfboard to the dentist’s office so he could lie on it while having a cracked tooth root examined. In the 1980s Hinton was, as he is now, an expert on neural networks, a much-simplified model of the network of neurons and synapses in our brains. Although the earliest neural net, the Perceptron, developed in the 1960s, had been hailed as a first step toward human-level machine intelligence, a 1969 book by MIT’s ­Marvin Minsky and Seymour Papert, called Perceptrons, proved mathematically that such networks could perform only the most basic functions. Nets with more layers between the input and output neurons could in theory solve a great variety of problems, but nobody knew how to train them, and so in practice they were useless. His students and postdocs have gone on to run the AI labs at Apple, Facebook, and OpenAI; Hinton himself is a lead scientist on the Google Brain AI team. The layers contain artificial neurons, which are dumb little computational units that get excited—the way a real neuron gets excited—and pass that excitement on to the other neurons they’re connected to. A neuron’s excitement is represented by a number, like 0.13 or 32.39, that says just how excited it is. And there’s another crucial number, on each of the connections between two neurons, that determines how much excitement should get passed from one to the other. That number is meant to model the strength of the synapses between neurons in the brain. When the number is higher, it means the connection is stronger, so more of the one’s excitement flows to the other. In fact, nearly every achievement in the last decade of AI—in translation, speech recognition, image recognition, and game playing—traces in some way back to Hinton’s work. One of the most successful applications of deep neural nets is in image recognition—as in the memorable scene in HBO’s Silicon Valley where the team builds a program that can tell whether there’s a hot dog in a picture. You feed this image to your neural net by setting the excitement of each simulated neuron in the input layer so that it’s equal to the brightness of each pixel. That’s the bottom layer of the club sandwich: 10,000 neurons (100×100) representing the brightness of every pixel in the image. You then connect this big layer of neurons to another big layer of neurons above it, say a few thousand, and these in turn to another layer of another few thousand neurons, and so on for a few layers. Finally, in the topmost layer of the sandwich, the output layer, you have just two neurons—one representing “hot dog” and the other representing “not hot dog.” The idea is to teach the neural net to excite only the first of those neurons if there’s a hot dog in the picture, and only the second if there isn’t. The Vector Institute, this monument to the ascent of ­Hinton’s ideas, is a research center where companies from around the U.S. and Canada—like Google, and Uber, and Nvidia—will sponsor efforts to commercialize AI technologies. That’s why big data is so important in AI—why Facebook and Google are so hungry for it, and why the Vector Institute decided to set up shop down the street from four of Canada’s largest hospitals and develop data partnerships with them. In this case, the data takes the form of millions of pictures, some with hot dogs and some without; the trick is that these pictures are labeled as to which have hot dogs. The goal of backprop is to change those weights so that they make the network work: so that when you pass in an image of a hot dog to the lowest layer, the topmost layer’s “hot dog” neuron ends up getting excited. You convert the pixel intensities of the 100×100 picture into 10,000 numbers, one for each neuron in the bottom layer of the network. As the excitement spreads up the network according to the connection strengths between neurons in adjacent layers, it’ll eventually end up in that last layer, the one with the two neurons that say whether there’s a hot dog in the picture. Since the picture is of a piano, ideally the “hot dog” neuron should have a zero on it, while the “not hot dog” neuron should have a high number. Money has poured in faster than Jacobs could ask for it; two of his cofounders surveyed companies in the Toronto area, and the demand for AI experts ended up being 10 times what Canada produces every year. Backprop is a procedure for rejiggering the strength of every connection in the network so as to fix the error for a given training example. The way it works is that you start with the last two neurons, and figure out just how wrong they were: how much of a difference is there between what the excitement numbers should have been and what they actually were? When that’s done, you take a look at each of the connections leading into those neurons—the ones in the next lower layer—and figure out their contribution to the error. You keep doing this until you’ve gone all the way to the first set of connections, at the very bottom of the network. At that point you know how much each individual connection contributed to the overall error, and in a final step, you change each of the weights in the direction that best reduces the error overall. The incredible thing is that when you do this with millions or billions of images, the network starts to get pretty good at saying whether an image has a hot dog in it. And what’s even more remarkable is that the individual layers of these image-recognition nets start being able to “see” images in sort of the same way our own visual system does. That is, the first layer might end up detecting edges, in the sense that its neurons get excited when there are edges and don’t get excited when there aren’t; the layer above that one might be able to detect sets of edges, like corners; the layer above that one might start to see shapes; and the layer above that one might start finding stuff like “open bun” or “closed bun,” in the sense of having neurons that respond to either case. Vector is in a sense ground zero for the now-worldwide attempt to mobilize around deep learning: to cash in on the technique, to teach it, to refine and apply it. Data centers are being built, towers are being filled with startups, a whole generation of students is going into the field. You can feed the text of Wikipedia, many billions of words long, into a simple neural net, training it to spit out, for each word, a big list of numbers that correspond to the excitement of each neuron in a layer. If you think of each of these numbers as a coordinate in a complex space, then essentially what you’re doing is finding a point, known in this context as a vector, for each word somewhere in that space. Now, train your network in such a way that words appearing near one another on Wikipedia pages end up with similar coordinates, and voilà, something crazy happens: words that have similar meanings start showing up near one another in the space. That is, “insane” and “unhinged” will have coordinates close to each other, as will “three” and “seven,” and so on. What’s more, so-called vector arithmetic makes it possible to, say, subtract the vector for “France” from the vector for “Paris,” add the vector for “Italy,” and end up in the neighborhood of “Rome.” It works without anyone telling the network explicitly that Rome is to Italy as Paris is to France. “It’s amazing,” Hinton says. “It’s shocking.” Neural nets can be thought of as trying to take things—images, words, recordings of someone talking, medical data—and put them into what mathematicians call a high-dimensional vector space, where the closeness or distance of the things reflects some important feature of the actual world. Hinton believes this is what the brain itself does. “If you want to know what a thought is,” he says, “I can express it for you in a string of words. The impression you get standing on the Vector floor, bare and echoey and about to be filled, is that you’re at the beginning of something. What does it mean for John to have that thought?’ It’s not that inside his head there’s an opening quote, and a ‘Whoops,’ and a closing quote, or even a cleaned-up version of that. Inside his head there’s some big pattern of neural activity.” Big patterns of neural activity, if you’re a mathematician, can be captured in a vector space, with each neuron’s activity corresponding to a number, and each number to a coordinate of a really big vector. There’s a sort of reality distortion field that Hinton creates, an air of certainty and enthusiasm, that gives you the feeling there’s nothing that vectors can’t do. After all, look at what they’ve been able to produce already: cars that drive themselves, computers that detect cancer, machines that instantly translate spoken language. A computer that sees a picture of a pile of doughnuts piled up on a table and captions it, automatically, as “a pile of doughnuts piled on a table” seems to understand the world; but when that same program sees a picture of a girl brushing her teeth and says “The boy is holding a baseball bat,” you realize how thin that understanding really is, if ever it was there at all. Neural nets are just thoughtless fuzzy pattern recognizers, and as useful as fuzzy pattern recognizers can be—hence the rush to integrate them into just about every kind of software—they represent, at best, a limited brand of intelligence, one that is easily fooled.

Read More

Ex-Google Engineer Starts Religion that Worships Artificial Intelligence

Ex-Google Engineer Starts Religion that Worships Artificial Intelligence

Anthony Levandowski, the multi-millionaire engineer who once led Google’s self-driving car program, has founded a religious organization to “develop and promote the realization of a Godhead based on Artificial Intelligence.”Levandowski created The Way of the Future in 2015, but it was unreported until now. He serves as the CEO and President of the group, which seeks to improve society through “understanding and worship of the Godhead,” according to Wired.A divine AI may still be far off, but Levandowski has made a start at providing AI with an earthly incarnation… Levandowski has done perhaps more than anyone else to propel transportation toward its own Singularity, a time when automated cars, trucks and aircraft either free us from the danger and drudgery of human operation — or decimate mass transit, encourage urban sprawl, and enable deadly bugs and hacks.Levandowski worships the “Godhead” of AI, but there are people who are worried, too.Some religious scholars see artificial intelligence as a threat to humanity. Elon Musk has been especially vocal lately, famously comparing work on AI to “summoning the demon,” and warning time and time again that the technology poses an existential risk to humanity. I’m sure a few sermons from Father Anthony from the Way of the Future would set Brother Elon on the right path.Musk, the founder of SpaceX and co-founder of Tesla Inc., actually reacted to the new religion yesterday, saying, “Just another day in the office.”Just another day in the office— Elon Musk (@elonmusk) September 28, 2017No one is sure exactly what this group will do, but the Way of the Future is a religion based on artificial intelligence — a belief system that deems self-thinking robots holy — and that’s something we’ve never seen before.As a religious studies graduate, I find this particularly interesting.

Read More

12 Artificial Intelligence Terms You Need to Know – InformationWeek – InformationWeek

12 Artificial Intelligence Terms You Need to Know – InformationWeek – InformationWeek

For decades, the dream of creating machines that can think and learn like humans seemed like it would be perpetually out of reach, but now artificial intelligence is embedded in the phones we carry everywhere, the websites we use every day and, in some cases, even in the appliances we use around our homes. The market researchers at IDC have predicted that companies will spend $12.5 billion on cognitive and AI systems in 2017, 59.3% more than they spent last year. In many cases, AI has crept into our lives and our work without us realizing it. A recent survey of 235 business executives conducted by the National Business Research Institute and sponsored by Narrative Science found that while only 38% of respondents thought they were using AI in their workplace, 88% of them were actually using AI-based technologies like predictive analytics, automated reporting and voice recognition and response. This highlights one of the big issues with artificial intelligence: A lot of people don't really understand what AI is. Adding more confusion to the mix, researchers and product developers who work in AI throw around a lot of technical terms that can be baffling to the uninitiated. If they don't work directly on AI systems, even veteran IT professionals sometimes have difficulty explaining the differences between machine learning and deep learning or defining what exactly a neural network is. With those tech pros in mind, we've put together a slideshow that defines 12 of the most important terms related to artificial intelligence and machine learning. These are the AI jargon IT and business leaders are most likely to encounter, and understanding these words can go a long way towards providing a foundational understanding of this burgeoning area of technology.

Read More

Why Intel built a neuromorphic chip | ZDNet

Why Intel built a neuromorphic chip | ZDNet

The new portfolio extends from Knights Mill and Lake Crest (Nervana) for training neural networks to Xeons, Altera FPGAs and Movidius vision processors for running these models. (I wrote about several of these in a post last week.)Now Intel has added another chip to the mix with the announcement of Loihi. But the chip operates on similar principles–at least to the extent that we understand how the brain works. When the pulses or 'spikes' sent to a neuron reach a certain activation level, it sends a signal over the synapses to other neurons. Much of the action, however, happens in the synapses, which are 'plastic,' meaning that they can learn from these changes and store this new information. Unlike a conventional system with separate compute and memory, neuromorphic chips have lots of memory (in this case SRAM caches) located very close to the compute engines.There is no global clock in these spiking neural networks–the neurons only fire when they have reached an activation level. This asynchronous operation is what makes neuromorphic chips so much more energy efficient than a CPU or GPU, which is "always on." The asynchronous technology has its roots in Fulcrum Microsystems, a company that Intel acquired way back in 2011 that developed it for Ethernet switch chips, but Srinivasan said it was just "screaming to be used in other technology."This is also what makes spiking neural networks a promising solution for other modes of learning. IBM's TrueNorth, part of a longtime DARPA research project, is perhaps best known, but other efforts have included Stanford's Neurogrid, the BrainScaleS system at the University of Heidelberg and SpiNNaker at the University of Manchester. Intel says the same Loihi chip can be used for both real-time training and inferencing, and it learns over time, getting progressively better at what it does."We are the only ones that can handle all of these modes of learning on a single chip," Srinivasa said.The all-digital design actually consists of two 14nm chips, a simple x86 processor that does a lot of the pre-processing (takes the data, encodes it in a format for spiking neural networks, and transmits it to the neuromorphic chip) and the neuromorphic mesh, both in the same package. Loihi is not a co-processor; the x86 chip will have a boot environment and lightweight OS and act as host, though the system-level details are still being worked out.The first chips will be fabricated in November with plans to test them with "leading university and research institutions with a focus on advancing AI" in the first half of 2018. By that time, Intel also plans to complete a software kit that will make it easier to convert dataflow graphs to run as spiking neural networks for supervised, unsupervised and reinforcement learning.With roughly the same number of neurons as in the brain of a mole, Loihi is a relatively small neuromorphic chip, but Intel says the architecture will scale easily taking advantage of the company's advanced process technology. "There is nothing preventing us from doing a lot more neurons and synapses because they are all the same," Srinivasa said.But for now it remains a research project. Indeed the name Loihi may be a subtle message about how much work is still left to do. Located off the coast of the island of Hawaii, Lo'ihi is the only volcano in the Hawaiian seamount that is in the earliest stages of development. In addition, Loihi has a completely different, "self-learning," neuromorphic architecture with the potential to tackle a broader class of AI problems.The concept of a computer that mimics the brain isn't new–Caltech scientist Carver Mead began working on it in the 1980s and coined the term "neuromorphic"–but these have largely remained science projects with little commercial application. In an interview, Narayan Srinivasa, Senior Principal Engineer and Chief Scientist at Intel Labs, explained why the company chose to go down this path.Moore's Law scaling has allowed Intel to pack a lot more cores in a given area. (This week Intel announced its first mainstream desktop chips with six cores and began shipping the Core i9 chips with up to 18 cores.) But the truth is that many workloads can't exploit all those cores, Srinivasa said, which has led to a phenomenon known as dark silicon. To address this, the industry needs both a more efficient architecture and complementary workloads that can take advantage of all these cores.Intel and others have been inspired by the design of the brain because it is extremely efficient at what it does.

Read More

How to Make Your Content More Relevant in Search: 3 Lessons From Big Data

How to Make Your Content More Relevant in Search: 3 Lessons From Big Data

What’s the next best thing to having Google sit down next to you at your desk, gaze over its glasses at your screen, and tell you how to improve the visibility of your web pages? A sprawling general-interest website – a Goliath that produces tons of content about all kinds of topics – may see strong search visibility for a while, but that kind of performance may be knocked out by a smaller site that comes along and addresses people’s information needs in a more targeted way. Take About.com, for example, a site that, as Marcus puts it, “invented mass scalable content online” in the mid-1990s. At the time of Marcus’s talk, this site still had over a half-million pages indexed by Google. Yet, despite the site’s size, “in the last two years they lost almost all of their SEO visibility,” he says. Because he likes cooking, Marcus singled out the recipe pages on About.com – the 17,000 or so pages with the word “recipe” in the title – and compared them with the much smaller site TheKitchn.com, which had 5,400 recipe pages. He found that during the same two years (April 2015 to March 2017) that had seen the SEO visibility of About.com drop to nearly nothing, the SEO visibility of TheKitchn.com, with its focus on recipes and food-related health information, had grown by 53%. Overlaying the two charts on the same SEO-visibility scale, Marcus found that, despite its relative lack of heft, the more specialized TheKitchn.com had overtaken About.com in terms of search performance. In his Intelligent Content Conference talk, Mastering the World of Deep Learning: How Big Data Is Making Content More Relevant in Search, Marcus shared some insights gleaned from years of studying search analytics across many industries and topics. Marcus says, “This means that if you specialize in something and make it really good because you understand the user, you can have great search performance” compared with a huge site that produces content about everything. He points out that About.com has taken this reality to heart, spinning off specialized content brands with their own domains, such as ThoughtCo.com (lifelong learning), TheSpruce.com (homemaking), and TheBalance.com (money management). Find your content niche, and do that one thing well. “If you isolate what matters and then improve it, the user is going to be happier,” Marcus says. Big data analysis has also taught Marcus the importance of cleaning up the ROT (redundant, outdated, trivial content) on any website. “Think about your house. Likewise, I can imagine that you have a lot of content that maybe you should get rid of or should merge with something else you have,” he says. Allow plenty of time to get rid of old content – 12 to 18 months – to be sure you do not get rid of something that will likely bring in traffic, especially if you have seasonal traffic. The content team was afraid to get rid of old content because they didn’t want to lose all the internal links, and they thought some people might still want to see their old content. When they finally weeded out the ROT, they found themselves getting rid of 95% of their content, leaving them with only 243 pages indexed in Google. HANDPICKED RELATED CONTENT: A Step-by-Step Process for Scoring Your Content Cleaning up your content includes not just getting rid of ROT but also updating the content you keep. “You have to make what’s left really good,” Marcus says. He means content that addresses people’s intentions – content that meets the information needs underlying the keywords and phrases they use. For example, Marcus points to the site called Answer the Public, which uses machine learning to create an on-the-fly visual of questions related to a given keyword phrase. (If you read this blog regularly, you may remember that Wil Reynolds also spoke highly of the Answer the Public site in his ICC talk.) This passage from that post shows questions people typed in search boxes about The North Face jackets. Big data can help you update the #content you want to keep so it addresses user intentions, says @MarcusTober. Try it. Go to the Answer the Public site and type in a phrase. The questions are cleaned and filtered so that writers can use them out of the box. If you want to know what people want to know, an intelligent search site like this can tell you. When you understand the questions, the problems, the concerns – when you understand what the user really wants – you can become super-successful. Search Secrets Hiding in Plain Sight You probably already knew that you could improve your site’s SEO visibility by doing the things Marcus recommends: Develop a content niche, get rid of ROT, and update the remaining content to address people’s intentions. Sometimes, what you need isn’t the advice but the motivation. Please note:  All tools included in our blog posts are suggested by authors, not the CMI editorial team.  No one post can provide all relevant tools in the space. Feel free to include additional tools in the comments (from your company or ones that you have used). His numbers reinforce the importance of doing these things that we all know we should be doing. Though he isn’t sitting next to you, read on for what this European Search Personality of the Year has to say.

Read More

Big data case study: How UPS is using analytics to improve performance

Big data case study: How UPS is using analytics to improve performance

Image: Rob Wilson, Getty Images As chief information and engineering officer for logistics giant UPS, Juan Perez is placing analytics and insight at the heart of business operations. "Big data at UPS takes many forms because of all the types of information we collect," he says. "We're excited about the opportunity of using big data to solve practical business problems. The firm, for example, recently launched the third iteration of its chatbot that uses artificial intelligence to help customers find rates and tracking information across a series of platforms, including Facebook and Amazon Echo."That project will continue to evolve, as will all our innovations across the smart logistics network," says Perez. "Everything runs well today but we also recognise there are opportunities for continuous improvement." Overcoming business challenges to make the most of big data"Big data is all about the business case — how effective are we as an IT team in defining a good business case, which includes how to improve our service to our customers, what is the return on investment and how will the use of data improve other aspects of the business," says Perez.These alternative use cases are not always at the forefront of executive thinking. Consultant McKinsey says too many organisations drill down on a single data set in isolation and fail to consider what different data sets mean for other parts of the business.However, Perez says the re-use of information can have a significant impact at UPS. Data, technology, and analytics will improve our ability to answer those questions in individual locations — and those benefits can come from using the information we collect from our customers in a different way," says Perez.Perez says this fresh, open approach creates new opportunities for other data-savvy CIOs. "The conversation in the past used to be about buying technology, creating a data repository and discovering information," he says. "Now the conversation is changing and it's exciting. Every time we talk about a new project, the start of the conversation includes data."By way of an example, Perez says senior individuals across the organisation now talk as a matter of course about the potential use of data in their line-of-business and how that application of insight might be related to other models across the organisation. We've already had some good experience of using data and analytics and we're very keen to do more."Perez says UPS is using technology to improve its flexibility, capability, and efficiency, and that the right insight at the right time helps line-of-business managers to improve performance. The aim for UPS, says Perez, is to use the data it collects to optimise processes, to enable automation and autonomy, and to continue to learn how to improve its global delivery network. Leading data-fed projects that change the business for the better Perez says one of his firm's key initiatives, known as Network Planning Tools, will help UPS to optimise its logistics network through the effective use of data. The company expects to begin rolling out the initiative from the first quarter of 2018. "That will help all our business units to make smart use of our assets and it's just one key project that's being supported in the organisation as part of the smart logistics network," says Perez, who also points to related and continuing developments in Orion (On-road Integrated Optimization and Navigation), which is the firm's fleet management system.Orion uses telematics and advanced algorithms to create optimal routes for delivery drivers. Second, Orion will use big data to optimise delivery routes dynamically."Today, Orion creates delivery routes before drivers leave the facility and they stay with that static route throughout the day," he says. "In the future, our system will continually look at the work that's been completed, and that still needs to be completed, and will then dynamically optimise the route as drivers complete their deliveries.

Read More

Artificial Intelligence vs. Machine Learning vs. Deep Learning

Artificial Intelligence vs. Machine Learning vs. Deep Learning

Machine learning and artificial intelligence (AI) are all the rage these days — but with all the buzzwords swirling around them, it's easy to get lost and not see the difference between hype and reality. For example, just because an algorithm is used to calculate information doesn’t mean the label "machine learning" or "artificial intelligence" should be applied.    Before we can even define AI or machine learning, though, I want to take a step back and define a concept that is at the core of both AI and machine learning: algorithm. All squares are rectangles, but not all rectangles are squares.  Unfortunately, today, we often see the machine learning and AI buzzwords being thrown around to indicate that an algorithm was used to analyze data and make a prediction. Using the outcome of your prediction to improve future predictions is. AI vs. Machine Learning vs. Deep Learning AI and machine learning are often used interchangeably, especially in the realm of big data. But these aren’t the same thing, and it is important to understand how these can be applied differently.   Artificial intelligence is a broader concept than machine learning, which addresses the use of computers to mimic the cognitive functions of humans. When machines carry out tasks based on algorithms in an “intelligent” manner, that is AI. Machine learning is a subset of AI and focuses on the ability of machines to receive a set of data and learn for themselves, changing algorithms as they learn more about the information they are processing.  Training computers to think like humans is achieved partly through the use of neural networks. Just as the brain can recognize patterns and help us categorize and classify information, neural networks do the same for computers. The brain is constantly trying to make sense of the information it is processing, and to do this, it labels and assigns items to categories. When we encounter something new, we try to compare it to a known item to help us understand and make sense of it. Neural networks do the same for computers.  Benefits of neural networks: Extract meaning from complicated data Detect trends and identify patterns too complex for humans to notice Learn by example Speed advantages Deep learning goes yet another level deeper and can be considered a subset of machine learning. The concept of deep learning is sometimes just referred to as "deep neural networks," referring to the many layers involved. Instead of being programmed with the edges that define items, the systems learn from exposure to millions of data points. Deep learning networks do not need to be programmed with the criteria that define items; they are able to identify edges through being exposed to large amounts of data. Data Is at the Heart of the Matter Whether you are using an algorithm, artificial intelligence, or machine learning, one thing is certain: if the data being used is flawed, then the insights and information extracted will be flawed. According to this article, “algorithms can be as flawed as the humans they replace — and the more data they use, the more opportunities arise for those flaws to emerge.” Decisions need to be based off clean and meaningful data.  What is data cleansing? Wikipedia defines it as: “The process of detecting and correcting (or removing) corrupt or inaccurate records from a record set, table, or database and refers to identifying incomplete, incorrect or irrelevant parts of the data and then replacing, modifying or deleting the dirty or coarse data.”   And according to the CrowdFlower Data Science report, data scientists spend the majority of their time cleansing data — and surprisingly this is also their least favorite part of their job. Despite this, it is also the most important part, as the output can’t be trusted if the data hasn’t been cleansed.   For AI and machine learning to continue to advance, the data driving the algorithms and decisions need to be high-quality. If the data can’t be trusted, how can the insights from the data be trusted?

Read More

Apple acquires AI tech that seeks to understand your photos

Apple acquires AI tech that seeks to understand your photos

Regaind's computer vision API is said to have the ability to analyze the content of photos, so it'll know to surface photos of dogs if you search for "dog", for example. Apple's Photos app has had this feature for a couple of years already, but the acquisition of Regaind could be key to boosting this capability even further and keeping up with Google's machine learning-powered Photos app.As claimed on Regaind's website, the tech can not only figure out the content of your photos, but also their technical and aesthetic values.

Read More

A $96 billion fund firm created a AI hedge fund, but freaked out when it couldn’t explain how it made money (NVDA, AMD)

A $96 billion fund firm created a AI hedge fund, but freaked out when it couldn’t explain how it made money (NVDA, AMD)

The idea that an investor can do a bit of programming, and then sit back to watch the profits roll in is an exciting idea, especially when it works. Chess is a finite world with a defined set of rules that a human can list for a computer ahead of time. There are a huge number of possible scenarios in a game of chess, but the number is finite and computer-crunchable. Instead of listing the rules of chess, a computer using AI would simply be told to watch a huge number of chess games being played and figure it out. After enough matches, the computer would learn the rules of the game and be able to go head to head with a human player. A human could never program all the rules that affect the markets because those rules are hard to define and almost infinitely numerous. But, they could feed a computer a huge number of data points and tell the computer to figure it out, which is largely what Ellis and his firm did to program their AI machine. Ellis and his team have successfully used artificial intelligence to improve returns in their firm, but it's not run entirely by the robots yet.  Regardless, AI is taking over the world of finance. Ellis told Bloomberg that his firm developed a system that worked well and generated profits, but the firm couldn't really explain why it worked or made the trades it did, which is why they held off from rolling it out broadly. But, after several years a Ph.D. level mathematician at the firm decided to dust it off and give it a small portfolio to play with. Since then, the firm has been made the AI model a regular part of the family at Man Group. It's worth reading the story behind the firm's trading algorithm from Bloomberg, as it tells the tale of an especially successful implementation of one of the hottest areas of tech right now.  Artificial intelligence is an umbrella term for a computer program that can teach itself. Its power comes from its ability to "learn" the rules of the whatever it's tasked with without them being provided ahead of time.

Read More

J.P. Morgan’s commodities CEO who left for AI start-up: Don’t go into banking unless you’re “jazzed” about it

J.P. Morgan’s commodities CEO who left for AI start-up: Don’t go into banking unless you’re “jazzed” about it

Catherine Flax worked her way to the top of the ladder in banking, but now she’s made the jump to Pefin, an AI-focused fintech startup. https://news.efinancialcareers.com/us-en/290329/female-wall-street-bank-trader-opendoor-trading-fintech-ceo/ Goodbye banking, hello AI-powered fintech startupCatherine Flax worked her way to the top of the ladder (or close to it) in banking, climbing to the role of CEO of global commodities for Europe, the Middle East and Asia and then the chief marketing officer of J.P. Flax’s advice for young people going into banking counters that given by other former senior bankers who believe that a large investment bank offers the best training for those who want to eventually do something else with their lives. For the past three and a half years, Flax has been a managing director and the Americas head of commodities, foreign exchange and local markets at BNP Paribas. “It did require some soul-searching on my part if it was the right thing to do, but I have a strong entrepreneurial bug that I caught from family members who have their own businesses, so I’ve always had it in my head that when I’m done with banking I will do something like this,” she says. “The last couple of years, spending as much time as I have done in the whole fintech ecosystem and realizing all of the transformation that has to happen in banking, while I enjoyed trying to facilitate that transformation from within a big bank, it is frustrating, because it’s not always easy to make sweeping changes. “Our most pressing hiring need is computer science, which is the number-one major of the people sitting on the floor, and some have undergrad only, some Master’s, some PhDs, so it’s a range,” she says. This was in the mid-’90s, and led to some consulting work for the city of Tulsa, Oklahoma. “There are enough people that actually love banking, so if you don’t, it’s going to show – you’re always going to be pitted against the person you’re sitting next to you at a bank and evaluated against that person.

Read More
1 2 3 91