Tag Archives: complexity
We’ve long used the brain as inspiration for computers, but the SpiNNaker supercomputer, switched on this month, is probably the closest we’ve come to recreating it in silicon. Now scientists hope to use the supercomputer to model the very thing that inspired its design.
The brain is the most complex machine in the known universe, but that complexity comes primarily from its architecture rather than the individual components that make it up. Its highly interconnected structure means that relatively simple messages exchanged between billions of individual neurons add up to carry out highly complex computations.
That’s the paradigm that has inspired the ‘Spiking Neural Network Architecture” (SpiNNaker) supercomputer at the University of Manchester in the UK. The project is the brainchild of Steve Furber, the designer of the original ARM processor. After a decade of development, a million-core version of the machine that will eventually be able to simulate up to a billion neurons was switched on earlier this month.
The idea of splitting computation into very small chunks and spreading them over many processors is already the leading approach to supercomputing. But even the most parallel systems require a lot of communication, and messages may have to pack in a lot of information, such as the task that needs to be completed or the data that needs to be processed.
In contrast, messages in the brain consist of simple electrochemical impulses, or spikes, passed between neurons, with information encoded primarily in the timing or rate of those spikes (which is more important is a topic of debate among neuroscientists). Each neuron is connected to thousands of others via synapses, and complex computation relies on how spikes cascade through these highly-connected networks.
The SpiNNaker machine attempts to replicate this using a model called Address Event Representation. Each of the million cores can simulate roughly a million synapses, so depending on the model, 1,000 neurons with 1,000 connections or 100 neurons with 10,000 connections. Information is encoded in the timing of spikes and the identity of the neuron sending them. When a neuron is activated it broadcasts a tiny packet of data that contains its address, and spike timing is implicitly conveyed.
By modeling their machine on the architecture of the brain, the researchers hope to be able to simulate more biological neurons in real time than any other machine on the planet. The project is funded by the European Human Brain Project, a ten-year science mega-project aimed at bringing together neuroscientists and computer scientists to understand the brain, and researchers will be able to apply for time on the machine to run their simulations.
Importantly, it’s possible to implement various different neuronal models on the machine. The operation of neurons involves a variety of complex biological processes, and it’s still unclear whether this complexity is an artefact of evolution or central to the brain’s ability to process information. The ability to simulate up to a billion simple neurons or millions of more complex ones on the same machine should help to slowly tease out the answer.
Even at a billion neurons, that still only represents about one percent of the human brain, so it’s still going to be limited to investigating isolated networks of neurons. But the previous 500,000-core machine has already been used to do useful simulations of the Basal Ganglia—an area affected in Parkinson’s disease—and an outer layer of the brain that processes sensory information.
The full-scale supercomputer will make it possible to study even larger networks previously out of reach, which could lead to breakthroughs in our understanding of both the healthy and unhealthy functioning of the brain.
And while neurological simulation is the main goal for the machine, it could also provide a useful research tool for roboticists. Previous research has already shown a small board of SpiNNaker chips can be used to control a simple wheeled robot, but Furber thinks the SpiNNaker supercomputer could also be used to run large-scale networks that can process sensory input and generate motor output in real time and at low power.
That low power operation is of particular promise for robotics. The brain is dramatically more power-efficient than conventional supercomputers, and by borrowing from its principles SpiNNaker has managed to capture some of that efficiency. That could be important for running mobile robotic platforms that need to carry their own juice around.
This ability to run complex neural networks at low power has been one of the main commercial drivers for so-called neuromorphic computing devices that are physically modeled on the brain, such as IBM’s TrueNorth chip and Intel’s Loihi. The hope is that complex artificial intelligence applications normally run in massive data centers could be run on edge devices like smartphones, cars, and robots.
But these devices, including SpiNNaker, operate very differently from the leading AI approaches, and its not clear how easy it would be to transfer between the two. The need to adopt an entirely new programming paradigm is likely to limit widespread adoption, and the lack of commercial traction for the aforementioned devices seems to back that up.
At the same time, though, this new paradigm could potentially lead to dramatic breakthroughs in massively parallel computing. SpiNNaker overturns many of the foundational principles of how supercomputers work that make it much more flexible and error-tolerant.
For now, the machine is likely to be firmly focused on accelerating our understanding of how the brain works. But its designers also hope those findings could in turn point the way to more efficient and powerful approaches to computing.
Image Credit: Adrian Grosu / Shutterstock.com Continue reading
Big data, personalized medicine, artificial intelligence. String these three buzzphrases together, and what do you have?
A system that may revolutionize the future of healthcare, by bringing sophisticated health data directly to patients for them to ponder, digest, and act upon—and potentially stop diseases in their tracks.
At Singularity University’s Exponential Medicine conference in San Diego this week, Dr. Ran Balicer, director of the Clalit Research Institute in Israel, painted a futuristic picture of how big data can merge with personalized healthcare into an app-based system in which the patient is in control.
Dr. Ran Balicer at Exponential Medicine
Picture this: instead of going to a physician with your ailments, your doctor calls you with some bad news: “Within six hours, you’re going to have a heart attack. So why don’t you come into the clinic and we can fix that.” Crisis averted.
Following the treatment, you’re at home monitoring your biomarkers, lab test results, and other health information through an app with a clean, beautiful user interface. Within the app, you can observe how various health-influencing life habits—smoking, drinking, insufficient sleep—influence your chance of future cardiovascular disease risks by toggling their levels up or down.
There’s more: you can also set a health goal within the app—for example, stop smoking—which automatically informs your physician. The app will then suggest pharmaceuticals to help you ditch the nicotine and automatically sends the prescription to your local drug store. You’ll also immediately find a list of nearby support groups that can help you reach your health goal.
With this hefty dose of AI, you’re in charge of your health—in fact, probably more so than under current healthcare systems.
Sound fantastical? In fact, this type of preemptive care is already being provided in some countries, including Israel, at a massive scale, said Balicer. By mining datasets with deep learning and other powerful AI tools, we can predict the future—and put it into the hands of patients.
The Israeli Advantage
In order to apply big data approaches to medicine, you first need a giant database.
Israel is ahead of the game in this regard. With decades of electronic health records aggregated within a central warehouse, Israel offers a wealth of health-related data on the scale of millions of people and billions of data points. The data is incredibly multiplex, covering lab tests, drugs, hospital admissions, medical procedures, and more.
One of Balicer’s early successes was an algorithm that predicts diabetes, which allowed the team to notify physicians to target their care. Clalit has also been busy digging into data that predicts winter pneumonia, osteoporosis, and a long list of other preventable diseases.
So far, Balicer’s predictive health system has only been tested on a pilot group of patients, but he is expecting to roll out the platform to all patients in the database in the next few months.
Truly Personalized Medicine
To Balicer, whatever a machine can do better, it should be welcomed to do. AI diagnosticians have already enjoyed plenty of successes—but their collaboration remains mostly with physicians, at a point in time when the patient is already ill.
A particularly powerful use of AI in medicine is to bring insights and trends directly to the patient, such that they can take control over their own health and medical care.
For example, take the problem of tailored drug dosing. Current drug doses are based on average results conducted during clinical trials—the dosing is not tailored for any specific patient’s genetic and health makeup. But what if a doctor had already seen millions of other patients similar to your case, and could generate dosing recommendations more relevant to you based on that particular group of patients?
Such personalized recommendations are beyond the ability of any single human doctor. But with the help of AI, which can quickly process massive datasets to find similarities, doctors may soon be able to prescribe individually-tailored medications.
Tailored treatment doesn’t stop there. Another issue with pharmaceuticals and treatment regimes is that they often come with side effects: potentially health-threatening reactions that may, or may not, happen to you based on your biometrics.
Back in 2017, the New England Journal of Medicine launched the SPRINT Data Analysis Challenge, which urged physicians and data analysts to identify novel clinical findings using shared clinical trial data.
Working with Dr. Noa Dagan at the Clalit Research Institute, Balicer and team developed an algorithm that recommends whether or not a patient receives a particularly intensive treatment regime for hypertension.
Rather than simply looking at one outcome—normalized blood pressure—the algorithm takes into account an individual’s specific characteristics, laying out the treatment’s predicted benefits and harms for a particular patient.
“We built thousands of models for each patient to comprehensively understand the impact of the treatment for the individual; for example, a reduced risk for stroke and cardiovascular-related deaths could be accompanied by an increase in serious renal failure,” said Balicer. “This approach allows a truly personalized balance—allowing patients and their physicians to ultimately decide if the risks of the treatment are worth the benefits.”
This is already personalized medicine at its finest. But Balicer didn’t stop there.
We are not the sum of our biologics and medical stats, he said. A truly personalized approach needs to take a patient’s needs and goals and the sacrifices and tradeoffs they’re willing to make into account, rather than having the physician make decisions for them.
Balicer’s preventative system adds this layer of complexity by giving weights to different outcomes based on patients’ input of their own health goals. Rather than blindly following big data, the system holistically integrates the patient’s opinion to make recommendations.
Balicer’s system is just one example of how AI can truly transform personalized health care. The next big challenge is to work with physicians to further optimize these systems, in a way that doctors can easily integrate them into their workflow and embrace the technology.
“Health systems will not be replaced by algorithms, rest assured,” concluded Balicer, “but health systems that don’t use algorithms will be replaced by those that do.”
Image Credit: Magic mine / Shutterstock.com Continue reading
It’s common to hear phrases like ‘machine learning’ and ‘artificial intelligence’ and believe that somehow, someone has managed to replicate a human mind inside a computer. This, of course, is untrue—but part of the reason this idea is so pervasive is because the metaphor of human learning and intelligence has been quite useful in explaining machine learning and artificial intelligence.
Indeed, some AI researchers maintain a close link with the neuroscience community, and inspiration runs in both directions. But the metaphor can be a hindrance to people trying to explain machine learning to those less familiar with it. One of the biggest risks of conflating human and machine intelligence is that we start to hand over too much agency to machines. For those of us working with software, it’s essential that we remember the agency is human—it’s humans who build these systems, after all.
It’s worth unpacking the key differences between machine and human intelligence. While there are certainly similarities, it’s by looking at what makes them different that we can better grasp how artificial intelligence works, and how we can build and use it effectively.
Central to the metaphor that links human and machine learning is the concept of a neural network. The biggest difference between a human brain and an artificial neural net is the sheer scale of the brain’s neural network. What’s crucial is that it’s not simply the number of neurons in the brain (which reach into the billions), but more precisely, the mind-boggling number of connections between them.
But the issue runs deeper than questions of scale. The human brain is qualitatively different from an artificial neural network for two other important reasons: the connections that power it are analogue, not digital, and the neurons themselves aren’t uniform (as they are in an artificial neural network).
This is why the brain is such a complex thing. Even the most complex artificial neural network, while often difficult to interpret and unpack, has an underlying architecture and principles guiding it (this is what we’re trying to do, so let’s construct the network like this…).
Intricate as they may be, neural networks in AIs are engineered with a specific outcome in mind. The human mind, however, doesn’t have the same degree of intentionality in its engineering. Yes, it should help us do all the things we need to do to stay alive, but it also allows us to think critically and creatively in a way that doesn’t need to be programmed.
The Beautiful Simplicity of AI
The fact that artificial intelligence systems are so much simpler than the human brain is, ironically, what enables AIs to deal with far greater computational complexity than we can.
Artificial neural networks can hold much more information and data than the human brain, largely due to the type of data that is stored and processed in a neural network. It is discrete and specific, like an entry on an excel spreadsheet.
In the human brain, data doesn’t have this same discrete quality. So while an artificial neural network can process very specific data at an incredible scale, it isn’t able to process information in the rich and multidimensional manner a human brain can. This is the key difference between an engineered system and the human mind.
Despite years of research, the human mind still remains somewhat opaque. This is because the analog synaptic connections between neurons are almost impenetrable to the digital connections within an artificial neural network.
Speed and Scale
Consider what this means in practice. The relative simplicity of an AI allows it to do a very complex task very well, and very quickly. A human brain simply can’t process data at scale and speed in the way AIs need to if they’re, say, translating speech to text, or processing a huge set of oncology reports.
Essential to the way AI works in both these contexts is that it breaks data and information down into tiny constituent parts. For example, it could break sounds down into phonetic text, which could then be translated into full sentences, or break images into pieces to understand the rules of how a huge set of them is composed.
Humans often do a similar thing, and this is the point at which machine learning is most like human learning; like algorithms, humans break data or information into smaller chunks in order to process it.
But there’s a reason for this similarity. This breakdown process is engineered into every neural network by a human engineer. What’s more, the way this process is designed will be down to the problem at hand. How an artificial intelligence system breaks down a data set is its own way of ‘understanding’ it.
Even while running a highly complex algorithm unsupervised, the parameters of how an AI learns—how it breaks data down in order to process it—are always set from the start.
Human Intelligence: Defining Problems
Human intelligence doesn’t have this set of limitations, which is what makes us so much more effective at problem-solving. It’s the human ability to ‘create’ problems that makes us so good at solving them. There’s an element of contextual understanding and decision-making in the way humans approach problems.
AIs might be able to unpack problems or find new ways into them, but they can’t define the problem they’re trying to solve.
Algorithmic insensitivity has come into focus in recent years, with an increasing number of scandals around bias in AI systems. Of course, this is caused by the biases of those making the algorithms, but underlines the point that algorithmic biases can only be identified by human intelligence.
Human and Artificial Intelligence Should Complement Each Other
We must remember that artificial intelligence and machine learning aren’t simply things that ‘exist’ that we can no longer control. They are built, engineered, and designed by us. This mindset puts us in control of the future, and makes algorithms even more elegant and remarkable.
Image Credit: Liu zishan/Shutterstock Continue reading
Disruptive business models are often powered by alternative financing. In Part 1 of this series, I discussed how mobile is redefining money and banking and shared some of the dramatic transformations in the global remittance infrastructure.
In this article, we’ll discuss:
AI financial advisors and robo traders
Let’s dive right back in…
Decentralized Lending = Democratized Access to Finances
Peer-to-peer (P2P) lending is an age-old practice, traditionally with high risk and extreme locality. Now, the P2P funding model is being digitized and delocalized, bringing lending online and across borders.
Zopa, the first official crowdlending platform, arrived in the United Kingdom in 2004. Since then, the consumer crowdlending platform has facilitated lending of over 3 billion euros ($3.5 billion USD) of loans.
Person-to-business crowdlending took off, again in the U.K., in 2005 with Funding Circle, now with over 5 billion euros (~5.8 billion USD) of capital loaned to small businesses around the world.
Crowdlending next took off in the US in 2006, with platforms like Prosper and Lending Club. The US crowdlending industry has boomed to $21 billion in loans, across 515,000 loans.
Let’s take a step back… to a time before banks, when lending took place between trusted neighbors in small villages across the globe. Lending started as peer-to-peer transactions.
As villages turned into towns, towns turned into cities, and cities turned into sprawling metropolises, neighborly trust and the ability to communicate across urban landscapes broke down. That’s where banks and other financial institutions came into play—to add trust back into the lending equation.
With crowdlending, we are evidently returning to this pre-centralized-banking model of loans, and moving away from cumbersome intermediaries (e.g. high fees, regulations, and extra complexity).
Fueled by the permeation of the internet, P2P lending took on a new form as ‘crowdlending’ in the early 2000s. Now, as blockchain and artificial intelligence arrive on the digital scene, P2P lending platforms are being overhauled with transparency, accountability, reliability, and immutability.
Artificial Intelligence Micro Lending & Credit Scores
We are beginning to augment our quantitative decision-making with neural networks processing borrowers’ financial data to determine their financial ‘fate’ (or, as some call it, your credit score). Companies like Smart Finance Group (backed by Kai Fu Lee and Sinovation Ventures) are using artificial intelligence to minimize default rates for tens of millions of microloans.
Smart Finance is fueled by users’ personal data, particularly smartphone data and usage behavior. Users are required to give Smart Finance access to their smartphone data, so that Smart Finance’s artificial intelligence engine can generate a credit score from the personal information.
The benefits of this AI-powered lending platform do not stop at increased loan payback rates; there’s a massive speed increase as well. Smart Finance loans are frequently approved in under eight seconds. As we’ve seen with other artificial intelligence disruptions, data is the new gold.
Digitizing access to P2P loans paves the way for billions of people currently without access to banking to leapfrog the centralized banking system, just as Africa bypassed landline phones and went straight to mobile. Leapfrogging centralized banking and the credit system is exactly what Smart Finance has done for hundreds of millions of people in China.
As artificial intelligence accesses even the most mundane mobile browsing data to assign credit scores, blockchain technologies, particularly immutable ledgers and smart contracts, are massive disruptors to the archaic banking system, building additional trust and transparency on top of current P2P lending models.
Immutable ledgers provide the necessary transparency for accurate credit and loan defaulting history. Smart contracts executed on these immutable ledgers bring the critical ability to digitally replace cumbersome, expensive third parties (like banks), allowing individual borrowers or businesses to directly connect with willing lenders.
Two of the leading blockchain platforms for P2P lending are ETHLend and SALT Lending.
ETHLend is an Ethereum-based decentralized application aiming to bring transparency and trust to P2P lending through Ethereum network smart contracts.
Secure Automated Lending Technology (SALT) allows cryptocurrency asset holders to use their digital assets as collateral for cash loans, without the need to liquidate their holdings, giving rise to a digital-asset-backed lending market.
While blockchain poses a threat to many of the large, centralized banking institutions, some are taking advantage of the new technology to optimize their internal lending, credit scoring, and collateral operations.
In March 2018, ING and Credit Suisse successfully exchanged 25 million euros using HQLA-X, a blockchain-based collateral lending platform.
HQLA-X runs on the R3 Corda blockchain, a platform designed specifically to help heritage financial and commerce institutions migrate away from their inefficient legacy financial infrastructure.
Blockchain and tokenization are going through their own fintech and regulation shakeup right now. In a future blog, I’ll discuss the various efforts to more readily assure smart contracts, and the disruptive business model of security tokens and the US Securities and Exchange Commission.
Parallels to the Global Abundance of Capital
The abundance of capital being created by the advent of P2P loans closely relates to the unprecedented global abundance of capital.
Initial coin offerings (ICOs) and crowdfunding are taking a strong stand in disrupting the $164 billion venture capital market. The total amount invested in ICOs has risen from $6.6 billion in 2017 to $7.15 billion USD in the first half of 2018. Crowdfunding helped projects raise more than $34 billion in 2017, with experts projecting that global crowdfunding investments will reach $300 billion by 2025.
In the last year alone, using ICOs, over a dozen projects have raised hundreds of millions of dollars in mere hours. Take Filecoin, for example, which raised $257 million in only 30 days; its first $135 million was raised in the first hour. Similarly, the Dragon Coin project (which itself is revolutionizing remittance in high-stakes casinos around the world) raised $320 million in its 30-day public ICO.
Some Important Takeaways…
Technology-backed fundraising and financial services are disrupting the world’s largest financial institutions. Anyone, anywhere, at anytime will be able to access the capital they need to pursue their idea.
The speed at which we can go from “I’ve got an idea” to “I run a billion-dollar company” is moving faster than ever.
Following Ray Kurzweil’s Law of Accelerating Returns, the rapid decrease in time to access capital is intimately linked (and greatly dependent on) a financial infrastructure (technology, institutions, platforms, and policies) that can adapt and evolve just as rapidly.
This new abundance of capital requires financial decision-making with ever-higher market prediction precision. That’s exactly where artificial intelligence is already playing a massive role.
Artificial Intelligence, Robo Traders, and Financial Advisors
On May 6, 2010, the Dow Jones Industrial Average suddenly collapsed by 998.5 points (equal to 8 percent, or $1 trillion). The crash lasted over 35 minutes and is now known as the ‘Flash Crash’. While no one knows the specific reason for this 2010 stock market anomaly, experts widely agree that the Flash Crash had to do with algorithmic trading.
With the ability to have instant, trillion-dollar market impacts, algorithmic trading and artificial intelligence are undoubtedly ingrained in how financial markets operate.
In 2017, CNBC.com estimated that 90 percent of daily trading volume in stock trading is done by machine algorithms, and only 10 percent is carried out directly by humans.
Artificial intelligence and financial management algorithms are not only available to top Wall Street players.
Robo-advisor financial management apps, like Wealthfront and Betterment, are rapidly permeating the global market. Wealthfront currently has $9.5 billion in assets under management, and Betterment has $10 billion.
Artificial intelligent financial agents are already helping financial institutions protect your money and fight fraud. A prime application for machine learning is in detecting anomalies in your spending and transaction habits, and flagging potentially fraudulent transactions.
As artificial intelligence continues to exponentially increase in power and capabilities, increasingly powerful trading and financial management bots will come online, finding massive new and previously lost streams of wealth.
How else are artificial intelligence and automation transforming finance?
Disruptive Remittance and Seamless Transactions
When was the last time you paid in cash at a toll booth? How about for a taxi ride?
EZ-Pass, the electronic tolling company implemented extensively on the East Coast, has done wonders to reduce traffic congestion and increase traffic flow.
Driving down I-95 on the East Coast of the United States, drivers rarely notice their financial transaction with the state’s tolling agencies. The transactions are seamless.
The Uber app enables me to travel without my wallet. I can forget about payment on my trip, free up my mental bandwidth and time for higher-priority tasks. The entire process is digitized and, by extension, automated and integrated into Uber’s platform (Note: This incredible convenience many times causes me to accidentally walk out of taxi cabs without paying!).
In January 2018, we saw the success of the first cutting-edge, AI-powered Amazon Go store open in Seattle, Washington. The store marked a new era in remittance and transactions. Gone are the days of carrying credit cards and cash, and gone are the cash registers. And now, on the heals of these early ‘beta-tests’, Amazon is considering opening as many as 3,000 of these cashierless stores by 2023.
Amazon Go stores use AI algorithms that watch various video feeds (from advanced cameras) throughout the store to identify who picks up groceries, exactly what products they select, and how much to charge that person when they walk out of the store. It’s a grab and go experience.
Let’s extrapolate the notion of seamless, integrated payment systems from Amazon Go and Uber’s removal of post-ride payment to the rest of our day-to-day experience.
Imagine this near future:
As you near the front door of your home, your AI assistant summons a self-driving Uber that takes you to the Hyperloop station (after all, you work in L.A. but live in San Francisco).
At the station, you board your pod, without noticing that your ticket purchase was settled via a wireless payment checkpoint.
After work, you stop at the Amazon Go and pick up dinner. Your virtual AI assistant passes your Amazon account information to the store’s payment checkpoint, as the store’s cameras and sensors track you, your cart and charge you auto-magically.
At home, unbeknownst to you, your AI has already restocked your fridge and pantry with whatever items you failed to pick up at the Amazon Go.
Once we remove the actively transacting aspect of finance, what else becomes possible?
Extraordinary transformations are happening in the finance world. We’ve only scratched the surface of the fintech revolution. All of these transformative financial technologies require high-fidelity assurance, robust insurance, and a mechanism for storing value.
I’ll dive into each of these other facets of financial services in future articles.
For now, thanks to coming global communication networks being deployed on 5G, Alphabet’s LUNE, SpaceX’s Starlink and OneWeb, by 2024, nearly all 8 billion people on Earth will be online.
Once connected, these new minds, entrepreneurs, and customers need access to money and financial services to meaningfully participate in the world economy.
By connecting lenders and borrowers around the globe, decentralized lending drives down global interest rates, increases global financial market participation, and enables economic opportunity to the billions of people who are about to come online.
We’re living in the most abundant time in human history, and fintech is just getting started.
Abundance Digital Online Community: I have created a Digital/Online community of bold, abundance-minded entrepreneurs called Abundance Digital. This is my ‘onramp’ for exponential entrepreneurs – those who want to get involved and play at a higher level. Click here to learn more.
Image Credit: Novikov Aleksey / Shutterstock.com Continue reading