MALWARE is a growing problem. According to KSN data and the Q1 2017 report, Kaspersky Lab solutions detected and repelled almost 500,00,000 malicious attacks from online resources located in 190 countries all over the world.
There's lots of talk about the application of Machine Learning in security. Matt Wolff, Chief Data Scientist at Cylance says that with a machine learning approach, many manual tasks performed by analysts can be automated, and even deployed in real time to catch these activities before any damage is done. He goes on to give the example of a well-trained machine learning model being able to identify unusual traffic on the network, and shut down these connections as they occur.
Think back to the last event you attended, or your experience at airport security when you last flew. - you will have queued for security screening. Machine Learning is being used to benefit this process by identifying risks that their human counterpart may missand the results are the speeding up of screening and improving safety.
CYBERCRIME is estimated to cost the global economy 400 billion dollars (source McAfee). Credit card fraud accounts for a large proportion of this cost.
While fraud detection is a mature process, the use of Machine learning / AI is very new. The industry now faces new challenges. Artificial Intelligence (AI) techniques are proposed to overcome the increasing challenges of online fraud. AI techniques are gaining popularity due to the power of Deep Learning Algorithms. Fraud prevention is a type of anomaly detection.
MACHINE-LEARNING is considered a subset of artificial intelligence (AI) that excels at finding patterns and making predictions and has been typically used by technology firms. Machine-learning is already used for tasks such as compliance, risk management and fraud prevention. The landscape however is changing as innovative fintech companies embrace the latest ML applications.
The newest use case for ML is in trading, where it is used both to crunch market data and to select and trade portfolios of securities. The Economist reported that Goldman has invested in Kensho, a startup that uses machine-learning to predict how events like natural disasters will affect market prices, based on data on similar events.
Natural Language Processing (NLP)
NATURAL-LANGUAGE PROCESSING is where AI-based systems are used on text.
Until now firms have used sophisticated analytics programmes to search for and reveal patterns hidden in structured data, such as spreadsheets and relational databases and then applied that data to improve their businesses. The structured data sources only account for about 20% of all available data
The exponential growth in data from the Internet, social media and personal devices- otherwise known as unstructured data and accounting for 80% of data, brings companies the opportunity incredible data insight. The real challenge was being able to mine this data that can’t be understood by computers. To extract value from unstructured data, companies across industries are turning to Natural Language Processing (NLP).
NLP enables computer programmes to interpret unstructured text by using machine learning and artificial intelligence to make inferences and provide context to language, just as human brains do.
Machine learning algorithms can be used in disease identification and diagnosis of ailments is at the forefront of ML research .Boston-based biopharma company Berg is using AI to research and develop diagnostics and therapeutic treatments in multiple areas, including oncology. ML algorithms can process more information and spot more patterns than their human counterparts. Examples include Google’s DeepMind Health, which last year announced multiple UK-based partnerships, including with Moorfields Eye Hospital in London, in which they’re developing technology to address macular degeneration in aging eyes..
THE EXPLOSION OF SOCIAL MEDIA sharing and the power of the influencer has changed consumer behaviour forever.
Gartner has reported 84 percent of millennials say User Generated Content (UGC) from strangers has some influence on what they buy.
Personalisation in marketing typically consisted of cookies, profiling, data analysis etc. Next-level personalisation means personalising customer experiences — leveraging UGC and machine learning to feature actual customer posts and photos within search results web pages.
Machine learning algorithms can then track the interactions consumers have with each piece of content to make intelligent recommendations based on consumers’ behavior. Using real content and intelligent technology means the future of marketing offers real life personalisation on a large scale for the first time ever.
E-COMMERCE companies use recommendation engines to persuade customers to buy additional products. If you're a regular shopper on amazon.com you'll have experienced this first hand, where Amazon’s recommendation engine gives you personalised shopping suggestions like "You might also like these...." or "People who bought this also bought..."
What about the science behind these recommendations? One of the most commonly used recommendation engines is collaborative filtering (CF) and its modifications where recommendation engines rely on likes and desires of other users in order to compute a similarity index between users and recommend items to them accordingly.
Google and its competitors are constantly improving what the search engine understands and finding uses for machine learning in its search algorithms.
Firms typically use query-based search to help consumers find information/products on their websites. the consideration is how to optimally rank a set of results shown in response to a query. Providing a personalised ranking based on a user’s search and click history is where ML can be applied.
Gary Illyes of Google says ML hasn't taken over the algorithm but they are typically used for coming up with new signals and signal aggregations.