Image

Machine Learning to Spot Fraudsters Applying for Credit

Callcredit helps businesses assess whether a person applying for credit is likely to pay it back through sophisticated analysis of a variety of datasets. It is using machine learning to understand whether a consumer is likely to default on a loan and spot fraudulent applications. It recently completed a trial of Microsoft Azure Machine Learning that suggested it could save credit card companies millions of pounds in bad debt. "Every data scientist that you talk to has a preference on a technique that they think is most optimal. We were determined to make sure that we found the best techniques for the right problem domains.", said Mark Davidson, Callcredit's Chief Data Officer. Most credit reference agencies rely on logistic regression models. Callcredit discovered that it was finding better predictions through boosted decision trees. Callcredit also tested how machine learning could improve fraud prevention. ML helped Callcredit identify conspicuous patterns in the application process that reveal whether a fraudster is likely applying for the debt facility. The company is now testing the scalability of the new methodology in a variety of use cases and building a decision-making scorecard for clients that will be released in the first quarter of 2018.

Image
Voice to Be the Future Source of Secure Authentication

If Artificial intelligence is to be believed, the human voice hints on information about your mood, social status, upbringing, age, ethnicity, weight, height, and facial features—plus information about the environment around you. This is where AI overtakes the capabilities of a human ear. “Voice profiling” is a concept where it can pick up micro signatures in a voice that reveal telling details about the speaker. Researchers at Carnegie Mellon University achieved a breakthrough by using artificial intelligence to generate a three-dimensional image of a speaker’s face, simply by analyzing a voice recording. The U.S. Coast Guard is already using Carnegie Mellon’s A.I. to build its own criminal cases against prank callers who send crews out on false search-and-rescue missions, which are costly and time-consuming.

Image
An AI Eye Scan Can Help to Diagnose Ailments

The light-sensitive layer found at the back of a person's eyes contains more than just cells that detect shadows and light — it also contains information about the health of a person's entire body. Google is using deep learning to predict a person’s blood pressure, age and smoking status by analyzing a photograph of their retina. Google believes that taking information from the arrangement of blood vessels can be used to predict whether someone is at risk of an impending heart attack. The technique relies on convolutional neural network, a type of deep-learning algorithm that is transforming how biologists analyze images. Convolutional neural networks allow computers to process an image efficiently and holistically, without splitting it into parts. Scientists also had to identify which types of study could be conducted using networks that must be trained with huge sets of images before they can start making predictions. When Google wanted to use deep learning to find mutations in genomes, its scientists had to convert strands of DNA letters into images that computers could recognize. Then they trained their network on DNA snippets that had been aligned with a reference genome, and whose mutations were known. The approach eliminates the need to stain cells—a process that requires more time and a sophisticated lab, and can damage the cell.

Image
Machine Learning Can Identify the Best Call Plan for Individual Customers

One major telecoms company Total Access Communication is focused on improving customer service through Artificial Intelligence and the Internet of Things. Total Access Communication Public Company Limited (or DTAC) has some 23.2 million subscribers and a market share of 30 percent. DTAC has a program of continuous improvement in place and this includes digital transformation initiatives. According to Piero Trivellato, head digital sales, “We’ve fully embraced their use for online marketing and have begun testing AI for customer care chatbots.” Chatbots lets customers to interact with services faster. Digital service also allows the telecoms operator to log calls and to analyze the results in a more streamlined fashion. Big data analytics can be used to improve services by understanding common customer queries. One application is used to assess which call package best suits an individual customer’s usage. A static algorithm is used. Over time, the machine learns which packages best match different customers. This eases the burden on call operators and reduces the complexity for customers in shifting through some twenty different call packages.

Image
AI To Detect Infected Machines Inside An Organization

Niddel, founded in 2014, hunts threats with machine learning, identifying potentially infected or compromised machines in an organization's network. It is being acquired by Verizon. Verizon wants to give its enterprise customers more tools to automate threat detections on networks. Niddle's main product is Magnet, that "investigates the relationship between indicators of compromise (IOCs), their inferred Tactics, Techniques and Procedures (TTPs) and log or event data generated by their organization," the company wrote on its website. Magnet transforms millions of raw data indicators into high confidence alerts, so analysts can focus on investigating qualified leads and eliminate repetitive tasks.

Image
AI Tool That Could Decode the Human Immune System

Microsoft has partnered with Adaptive Biotechnologies to map the genetics of the human immune system, to improve diagnoses of cancers and other diseases. Microsoft will do research and large-scale machine learning to translate genetics of the human immune system, or immunome into simple blood test that can be broadly accessible to people around the world. “We are very excited and inspired by our collaboration with Adaptive Biotechnologies, as it clearly advances our mission to use cloud and AI technologies to transform healthcare and improve the lives of people around the world,” said Peter Lee, VP, AI and Research, Microsoft. Adaptive, which was formed in 2009 with technology from the Fred Hutchinson Cancer Research Center, works to understand the immune system’s response to various diseases. The company is developing a map of sorts to track immune responses. Immune system is nature's most finely-tuned diagnostic as it routinely scans and reads any signal of disease - such as a cancer cell or an infectious agent - in the body, and holds the genetic code that can give insights into detecting these diseases. Collaboration is a part of Microsoft's Healthcare NExT initiative to accelerate innovation in the healthcare industry.

Image
AI Helps to Make Loans Through Social Media

MyBucks has been using artificial intelligence to help calculate credit scores and make loan decisions and has even begun offering 15-minute, AI-based loans through WhatsApp and Facebook Messenger. MyBucks is a Luxembourg-based fintech that provides loans and basic banking products. MyBucks’ app called “Haraka” can score a customer within two minutes. Customer downloads the app from the Android store and MyBucks pays for that download through reverse billing. So, customers who are broke can still access the app. The app then scrapes the phone for data, including text messages, call-history patterns and geolocation information. Information like the text messages of heavy users of mobile payment programs like M-Pesa contain all their mobile money transaction verifications. This information is very useful. “That gives us insight into customers’ income and expenses,” said Tim Nuy, deputy CEO at MyBucks. Customers also log into their social media accounts from the app to help MyBucks verify their identities: The company compares the applicant’s social media feed against the information in their mobile wallet. The bank starts with very small loans. A first loan to new customers might be a mere ten euros. If such borrowers successfully repay, they will qualify for a larger loan the second time. The returns made on those repeater loans make up for the slightly higher default rate — around 20% — on the first loans. MyBucks began offering loans through a chatbot in WhatsApp and Facebook Messenger in October.

Image
Google Chrome to Use Machine Learning to Protect Users from Malicious Extensions

Google has introduced a new method to detect confusing and deceptive browser extensions on Chrome. Soon Google will start using machine learning as an expansion of abuse protection to reduce harm to Chrome users. It will now incorporate machine learning to look at each inline installation request for bad signals in ads and web pages. Once Chrome detects the signals, it will selectively disable the request and redirect users to the extension page on the Web Store. Chrome already has an extension-level protection. However, these few extensions generate 90 percent more user complaints on an average than the remaining extensions on the Chrome Web Store. The automated enforcement system is in place to be responsive to user feedback, Google says.

Image
Scientists Want to Deal with Blindness Using Machine Learning

The human brain relies on the sense of touch, among other senses, to understand the physical world around us. One can distinguish between an apple and an orange by touching it. A group of scientists from Kolkata have leveraged this ability to train computers into decoding what one person is reading by its touch. Scientists have created a proof of principle brain-computer interface (BCI) that can identify 3D text as it is being touched by healthy individuals. Feel of series of embossed letters, numbers or symbols, a computer can discern the text just by reading the electric signals shooting through your brain. The goal of such scientific advances is to create devices that will respond directly to the brain instead of having signals relayed through your body. Rohit Bose from BCIs examined brain activity while healthy blindfolded people touched and recognized 3D letters, numbers and symbols. After participants touched the text, the computer was asked to identify the character being felt. When the experiment was repeated with different combinations one after another, both the accuracy of the prediction and the processing time took a hit. Accuracy declined to 65% and the time for prediction took up to two seconds. The paper was published in the journal Cognitive Neurodynamics on September 6, 2017.

Image

Disclaimer

Image