How Data Science Is Helping in Robotics and Artificial Intelligence

Big data and data science are set to bring in a digital revolution with groundbreaking technologies like artificial intelligence (AI), machine learning (ML), and deep learning. The essence of data science is to dive into massive datasets to extract meaningful information from them. The insights that data scientists and data analysts obtain from large volumes of data is the secret sauce that’s rapidly transforming everything around us. Institutions and organizations across various sectors of the industry are now leveraging data science technologies to power innovation and technology-driven change. In fact, nearly 53 percent of companies have adopted big data analytics in 2017, which is an enormous growth from the 17 percent in 2015.

As more and more companies are inclining towards data science to transform their organizational infrastructure for the better, it is giving rise to new and exciting career opportunities such as data scientists, data analysts, ML engineers, data architects, big data engineers, and so on. Thus, if you wish to start a career in data science, the time is now. There are plenty of resources available today to help you get started in data science and online platforms offering specialized data science courses are an excellent option. The advantage of online courses is that you can master data science concepts at your own pace and convenience.

Data Science and Artificial Intelligence

The fields of data science, AI, and ML are intrinsically linked to one another. While artificial intelligence is a broad umbrella that includes a wide range of applications, right from text analysis to robotics, machine learning, is a subset of artificial intelligence that focuses on training machines how to ‘learn’ via advanced algorithms and perform specific tasks while simultaneously improving performance through experience. Data science is a branch of computer science that deals in the extraction of valuable insights from vast datasets through a combination of disciplines such as mathematics, ML, statistics, and data engineering.

Today, AI and ML technologies are transforming the industrial landscape, and this is possible only because these technologies are backed by data science. While AI is about creating “intelligent and smart” machines, it cannot do without ML. As mentioned before, machine learning algorithms are required to train machines to learn from behavior patterns and cues. Then again, ML cannot function without analytics, which in turn cannot function without data infrastructure. Harvard Business Review maintains that, companies with strong basic analytics — such as sales data and market trends — make breakthroughs in complex and critical areas after layering in artificial intelligence.” However, for AI to create a true impact, you require the right data and a team of experienced and trained data science professionals who know where to look for the data and how to integrate it with AI and ML tools.

Let’s take the example of smart personal assistants like Siri, Alexa, and Cortana. These smart assistants represent the ‘Narrow AI’ and can interact with you and perform a limited number of tasks such as play songs for you, tell you about the day’s weather, or even do a little shopping for you. But, as we said, they can only perform ‘limited’ tasks since they have been exclusively ‘trained’ to do so. As data science continues to evolve in the future, data scientists might be able to tweak the algorithms of these assistants into more advanced ones (General AI) and then, maybe intelligent assistants can perform more complex tasks with much more precision than humans.

Data Science and Robotics

With the advance in data science, the field of robotics has definitely improved to a great extent. During the initial days of development, scientists were faced with two major challenges -one, predicting every action of a robot, and two, reducing the computational complexity in real-time vision tasks.

While robots could perform specific functions, it was impossible for scientists to predict their next move. For every new functionality, a robot would have to be reprogrammed every time, which made the task a tedious one. Another major obstacle with robots is that unlike humans who use their unique sense of vision to make sense of the world around them, robots can only visualize the world in a series of zeros and ones. Thus, accomplishing real-time vision tasks for robots would mean a fresh set of zeros and ones every time a new trend emerges, thereby increasing the computational complexity.

Enter machine learning to solve these issues in robotics. With ML, robots can acquire new behavior patterns through labeled data. Handwriting recognition is an excellent example. In handwriting recognition, computers are fed with labeled data — both positive and negative. Once the computer has successfully learned to differentiate between positive and negative examples, it is presented with new data. Based on the previous experience (during the training phase), the computer can predict the qualified classifiers for recognizing the handwriting. Thanks to advanced ML algorithms powered by tons of data that computers are now able to perform handwriting recognition much more accurately than they were ten years ago.

Furthermore, reinforcement learning, the branch of ML that is “the closest that machine learning can get to the way how humans learn” teaches computers and robots to perform specific functions according to their environment to generate outcomes that fetch either rewards or penalties. Thus, every time the robots lead to penalties, they can learn from their mistakes and know what course of action to take to fetch awards. Personalized recommendation lists of online portals such as Amazon and Netflix are the best examples of reinforcement learning. This wasn’t possible ten years ago!

As data scientists continue to leverage AI and ML to develop smart machines, in the process, they are gaining a deeper insight into the world of data science itself. Using AI and ML, data scientists and analysts can process, analyze, and interpret vast datasets much faster than ever. For instance, the MIT Data Science Machine can process large volumes of data and produce better predictive models anywhere between two to twelve hours, while the same would take months if done manually by data scientists. Another excellent case in point is that of California’s NuMedii Labs. Data scientists at the NuMedii Labs used network-based data mining algorithms to identify the correlations between the disease information and the drug composition to estimate the drug efficacy accurately. In this way, NuMedii aims to reduce the amount of time and risk associated with the process of drug development by bringing effective drugs into the market much faster than would happen through traditional methods.

Thus, data science, AI, and robotics have a pretty much symbiotic relationship. Each enhances the other to power innovative machines and technologies that are making our lives more convenient than ever. The collaboration between data science, AI, and ML has given us things like self-driving cars, smart assistants, robo-surgeons and nurses, and so much more. In the future, more is to come!

UCLA-developed artificial intelligence device identifies objects at the speed of light

The 3D-printed artificial neural network can be used in medicine, robotics and security

IMAGE
The network, composed of a series of polymer layers, works using light that travels through it. Each layer is 8 centimeters square.

A team of UCLA electrical and computer engineers has created a physical artificial neural network — a device modeled on how the human brain works — that can analyze large volumes of data and identify objects at the actual speed of light. The device was created using a 3D printer at the UCLA Samueli School of Engineering.

Numerous devices in everyday life today use computerized cameras to identify objects — think of automated teller machines that can “read” handwritten dollar amounts when you deposit a check, or internet search engines that can quickly match photos to other similar images in their databases. But those systems rely on a piece of equipment to image the object, first by “seeing” it with a camera or optical sensor, then processing what it sees into data, and finally using computing programs to figure out what it is.

The UCLA-developed device gets a head start. Called a “diffractive deep neural network,” it uses the light bouncing from the object itself to identify that object in as little time as it would take for a computer to simply “see” the object. The UCLA device does not need advanced computing programs to process an image of the object and decide what the object is after its optical sensors pick it up. And no energy is consumed to run the device because it only uses diffraction of light.

New technologies based on the device could be used to speed up data-intensive tasks that involve sorting and identifying objects. For example, a driverless car using the technology could react instantaneously — even faster than it does using current technology — to a stop sign. With a device based on the UCLA system, the car would “read” the sign as soon as the light from the sign hits it, as opposed to having to “wait” for the car’s camera to image the object and then use its computers to figure out what the object is.

Technology based on the invention could also be used in microscopic imaging and medicine, for example, to sort through millions of cells for signs of disease.

The study was published online in Science on July 26.

“This work opens up fundamentally new opportunities to use an artificial intelligence-based passive device to instantaneously analyze data, images and classify objects,” said Aydogan Ozcan, the study’s principal investigator and the UCLA Chancellor’s Professor of Electrical and Computer Engineering. “This optical artificial neural network device is intuitively modeled on how the brain processes information. It could be scaled up to enable new camera designs and unique optical components that work passively in medical technologies, robotics, security or any application where image and video data are essential.”

The process of creating the artificial neural network began with a computer-simulated design. Then, the researchers used a 3D printer to create very thin, 8 centimeter-square polymer wafers. Each wafer has uneven surfaces, which help diffract light coming from the object in different directions. The layers look opaque to the eye but submillimeter-wavelength terahertz frequencies of light used in the experiments can travel through them. And each layer is composed of tens of thousands of artificial neurons — in this case, tiny pixels that the light travels through.

Together, a series of pixelated layers functions as an “optical network” that shapes how incoming light from the object travels through them. The network identifies an object because the light coming from the object is mostly diffracted toward a single pixel that is assigned to that type of object.

The researchers then trained the network using a computer to identify the objects in front of it by learning the pattern of diffracted light each object produces as the light from that object passes through the device. The “training” used a branch of artificial intelligence called deep learning, in which machines “learn” through repetition and over time as patterns emerge.

“This is intuitively like a very complex maze of glass and mirrors,” Ozcan said. “The light enters a diffractive network and bounces around the maze until it exits. The system determines what the object is by where most of the light ends up exiting.”

In their experiments, the researchers demonstrated that the device could accurately identify handwritten numbers and items of clothing — both of which are commonly used tests in artificial intelligence studies. To do that, they placed images in front of a terahertz light source and let the device “see” those images through optical diffraction.

They also trained the device to act as a lens that projects the image of an object placed in front of the optical network to the other side of it — much like how a typical camera lens works, but using artificial intelligence instead of physics.

Because its components can be created by a 3D printer, the artificial neural network can be made with larger and additional layers, resulting in a device with hundreds of millions of artificial neurons. Those bigger devices could identify many more objects at the same time or perform more complex data analysis. And the components can be made inexpensively — the device created by the UCLA team could be reproduced for less than $50.

While the study used light in the terahertz frequencies, Ozcan said it would also be possible to create neural networks that use visible, infrared or other frequencies of light. A network could also be made using lithography or other printing techniques, he said.

###

The study’s others authors, all from UCLA Samueli, are postdoctoral scholars Xing Lin, Yair Rivenson, and Nezih Yardimci; graduate students Muhammed Veli and Yi Luo; and Mona Jarrahi, UCLA professor of electrical and computer engineering.

The research was supported by the National Science Foundation and the Howard Hughes Medical Institute. Ozcan also has UCLA faculty appointments in bioengineering and in surgery at the David Geffen School of Medicine at UCLA. He is the associate director of the UCLA California NanoSystems Institute and an HHMI professor.

Facebook asks banks for YOUR account details

Facebook has not had a good year. First there was the Cambridge Analytica scandal and then its share price fell like a stone from a skyscraper. And yet, just a few days ago The Wall Street Journal reported that the social media megalith is asking major US banks to share detailed financial information about how customers spend their money, apparently to increase user engagement. What could possibly go wrong, as we like to say when we can see all kinds of catastrophes likely to emerge from what is made to appear simple and innocuous?

Hand over your data!

The banks it has approached are well known to all, even if we don’t live in the United States. They are Wells Fargo, JP Morgan Chase and Citigroup. Facebook has, according to the WSJ’s interview with banking insiders, asked them to hand over data, including “everything from customers’ account balances to their credit card transactions.”

What’s in it for the banks?

Why might the banks agree to collaborate? The answer lies in the mobile commerce where apps like PayPal and Venmo dominate the scene. The banks would like a slice of this action and Facebook could help them achieve it. The article reports that Facebook is offering the banks “a presence on its Messenger app”, which has around 1.3 billion users. Messenger users can send and receive money via the app, BUT, at the moment, if a user wants to connect the Messenger app with their bank account they have to ‘opt-in’ to do that. They can also use the app to get in direct contact with Facebook’s credit card partners. The suggestion is that Messenger could also offer the same kind of direct contact with the banks.

Trapped by Facebook’s Messenger app

It’s easy to see the benefit to the banks and to Facebook, which is hoping that such a service will mean users conduct all their financial transactions through Messenger. Apparently the fact that many users leave the app to go and check their account balances at their bank’s online service is annoying Zuckerberg & Co who would prefer its users never wander off. Facebook also promises that it would never use customers’ financial data to improve its ad targeting. There are probably a lot of people reading that and thinking, “If you believe that, you’ll believe anything.”

So far, the banks have declined Facebook’s offer, citing customer privacy as a concern. And they are right to be concerned about it, because Facebook’s track record on use of customer data is covered in mud and it has stuck.

Knowing what you know about Facebook, how would you feel about your bank handing over your data to such a company?

Do You Know The Difference Between Data Analytics And AI Machine Learning?

The artificial intelligence (AI) industry has been leading the headlines consistently, and for good reason. It has already transformed industries across the globe, and companies are racing to understand how to integrate this emerging technology.

Artificial intelligence is not a new concept. The technology has been with us for a long time, but what has changed in recent years is the power of computing, cloud-based service options and the applicability of AI to our jobs as marketers.

AI’s impact on marketing is growing, predicted to reach nearly $40 billion by 2025. Most CMOs are aware of AI, but many are still unsure and unaware of the magnitude of the benefits and how they can adopt AI to improve marketing.

Advances in AI now mean product developers can create innovative and leading-edge products and services that, until recently, would not have been within reach of the average marketing budget.

These new products and services entering the market make AI adoption lower risk with a focus on delivering practical and immediately impactful results. Many past attempts resulted in expensive and custom-developed marketing technology projects that left their scars.

But before marketers commit to and execute their AI strategy, they need to understand the opportunity and difference between data analytics, predictive analytics and AI machine learning.

Data Analytics

Marketing managers have readily engaged with data analytics, benefitting (and most likely suffering) from the mountains of data at their fingertips. This includes everything from user-tracking data on apps and websites, newsletter conversion rates and online advertising click-throughs, to CRM data analysis.

Data mining delivers vast quantities of data, often unstructured. Marketers are more familiar with interacting with data via dashboards that structure data to deliver analysis of commonalities, such as averages, ratios and percentages. The goal is to aggregate data in order to report a result, search for a pattern and find relationships between variables. Assumptions are made by humans, and data is queried to attest to that relationship. If valid, testing may continue on additional data.

For the sake of example, let’s say that AnyBank credit card loyalty program uses data analytics to determine that it has 10,000 middle-aged male members, and 1,000 of them have redeemed their accumulated points for golf. The data helps make the assumption that middle-aged males are more likely to golf, and therefore AnyBank’s marketing efforts focus on this segment.

Data analysis is descriptive since it is based on past events. It does not predict the impact of a change in a variable.

Predictive Analytics

Data analytics leads naturally to predictive analytics using collected data to predict what might happen. Predictions are based on historical data and rely on human interaction to query data, validate patterns, create and then test assumptions.

Assumptions drawn from past experiences presuppose the future will follow the same patterns. “What/if” assumptions are informed by human understanding of the past, and predictive capability is limited by the volume, time and cost constraints of human data analysts.

Going back to our earlier example, AnyBank’s credit card loyalty program might use predictive analytics to determine whether they could increase reward redemption by 20% by spending 10% more on advertising golf to middle-aged male members. AnyBank could make the assumption that middle-aged males like to golf, so it markets to this segment and predicts that, based on past redemption rates from other specials, they will increase redemptions in-line with that result.

Predictive insights derived from data analytics are extremely useful to marketers. They can help predict campaign effectiveness, inform decision-making on collateral, geographic markets and demographics to target. But the more detailed the desire to target and segment, the higher the time and cost demands, making successful, hyper-personalized campaigning nearly impossible.

AI Machine Learning

Machine learning is a continuation of the concepts around predictive analytics, with one key difference: The AI system is able to make assumptions, test and learn autonomously.

AI is a combination of technologies, and machine learning is one of the most prominent techniques utilized for hyper-personalized marketing. AI machine learning makes assumptions, reassesses the model and reevaluates the data, all without the intervention of a human. This changes everything. Just as AI means that a human engineer does not need to code for each and every possible action/reaction, AI machine learning is able to test and retest data to predict every possible customer-product match, at a speed and capability no human could attain.

For example, AnyBank’s credit card loyalty program could utilize machine learning to determine that 1,000 of its male members live near a golf course, have not golfed before but enjoy sports. It also determines that many women members in the loyalty program are equally likely to be interested. It also sets parameters for the golf season in certain climate zones, such as the Southern U.S. It further determines a microsegment to offer Saturday afternoon to men and women without young children, who can more likely take the time on a Saturday. The system also acknowledges other sports-loving middle-aged men and women who live near a golf course and have young children, which would prompt other, more family-oriented offers.

Complex analysis, such as the example above, can be done instantaneously with many more variables involved, allowing the system to rapidly learn. This learning can deliver microtarget insights that could not be realistically done by human analysts across a large population. These results can dramatically improve conversion rates, marketing return on investment and customer loyalty.

It’s not a matter of one or the other — it is imperative that marketers understand the benefits and limitations of each. Simplified down:

• Data analysis refers to reviewing data from past events for patterns.

• Predictive analytics is making assumptions and testing based on past data to predict future what/ifs.

• AI machine learning analyzes data, makes assumptions, learns and provides predictions at a scale and depth of detail impossible for individual human analysts.

CMOs are increasingly required to make decisions that have significant technology implications. Understanding the difference between data analytics and AI is all about choosing the right tools for the right job.

This article was originally posted on https://www.forbes.com/sites/forbesagencycouncil/2018/08/01/do-you-know-the-difference-between-data-analytics-and-ai-machine-learning/#2c0d4d435878