Japan’s Resona Bank has partnered with data analysis expert Teradata, to build a next-generation marketing engine. It aims to provide customer with information and services tailored to their specific needs. The system analyzes the data of each customer and provides information and proposals according to the individual customer data anytime and anywhere. It will help bank to optimize its business operations via the use of AI technology and developing a sophisticated customer analysis by using digital data. The analytic solution has developed capabilities to predict behavioral paths of each individual customer, determining the next best interaction and delivering a consistent, personalized brand experience. The result is that it enables marketers to optimize objectives such as response and conversion rates, service delivery, churn, and customer satisfaction.
Artificial Intelligence to Screen Applicants For HR
Employers are looking at artificial intelligence to help find the best candidates for the job. This will speed up the entire process and saves a lot many man-hours of the teams across organizations. HireVue Assessments, a video interview platform, uses predictive analytics to find their suitable candidate. “We use the science of industrial and organizational psychology and the technology of artificial intelligence to analyze the input that we get from a video interview, and we use that to rank where an individual might fit for a potential role,” said Diana Kucer, the chief marketing officer for HireVue. The algorithm considers verbal and non-verbal clues, along with other parameters like facial expressions, body language, word choice etc. to short list its prospective hires. HR teams in organization like Children’s Mercy Hospital have already been using a version of the software for seven years.
Artificial Intelligence to Decide Prison-Sentence for The Accused
In US, across multiple cities, a “risk assessment algorithm” is assisting a magistrate decide whether an accused be granted release or booked for their crimes. The algorithm is trained on thousands of criminal records and weighing anywhere from a handful to dozens of factors, it would give out a recommendation for the judge to consider. But researchers are questioning its algorithm now. A new study suggests that a common risk assessment algorithm is just as accurate as a random person paid a dollar to guess whether or not someone will be arrested again. Furthermore, studies have shown that algorithms trained on racist data have big error rates for communities of colour. Another independent study by Megan Stevenson suggests that whatever algorithms predict, there is no guarantee judges or bail decision-makers will use their forecasts in a way that consistently reduces incarceration or protects public safety.
Teaching Artificial Intelligence to Becomes Its Own Teacher
DeepMind, a Google-owned company that had developed AlphaGo, defeated the European Go champion Fan Hui. Six months later, Google invited the world’s finest Go player, grandmaster Lee Sedol of South Korea, to a game of five matches. A documentary titled AlphaGo, available of Netflix, goes into how the company developed this program and how does one program a program to teach itself. Researchers created AlphaGo after showing it hundreds of thousands of games. They also made it play against different versions of itself thousands of times. AlphaGo Zero, the next generation, had learnt to play the game without ever watching a human play. It learned by playing millions of games against itself. AlphaGo Zero had learnt Go from scratch. The programmers just fed in the rules of the game; the rest it learnt for itself.
AI to Predict Vehicles Trajectory on Highways
In an attempt , which can be labelled as premature currently, to enable fully autonomous vehicles make consistent trajectory prediction, Florent Altché and Arnaud de La Fortelle have come up with a research paper, which can be labelled as premature currently. Currently, most advanced Driving Assistance Systems (ADAS) are unable to predict medium-term trajectories for the vehicles. Algorithms are limited only to response with high-likelihood situations such as emergency braking. In the paper they suggest solution by introducing a long short-term memory (LSTM) neural network which is capable of accurately predicting the future longitude and trajectories of its vehicles. They algorithm is trained on NGSIM US- 101 dataset, which contains a total of 800 hours of recorded trajectories in various traffic densities, representing more than 6000 individual drivers. The research showed to achieve better prediction accuracy than the previous benchmark. The use of the actual predictions combined with ADAS is a critical development for safe and efficient autonomous driving in future.
Storytelling Through Artificial Intelligence
With AI making way into almost every corner of our lives, storytelling through robots powered by AI isn’t a distance reality. The Hemingway App, an online writing editor, uses natural language processing (NLP) to recognize common writing problems and increase readability. It can also re-write an entire paragraph completely on its own. There are other possible options available too. “Dragon Diction” and “Dictation.io” can help one to write without typing. The algorithm of this AI powered app is based upon sentiment-based text analysis. It was trained on legendary author Kurt Vonnegut's lectures on shapes of stories. As per research, most popular novels are those based around the rise-fall-rise pattern of Cinderella followed by a tragedy, and a fall-rise-fall-rise pattern. AI can also easily inspect how many comments there are on an online video and digest all kinds of data about how well it performs on social media. Armed with such skills and this data, any computer can construct a basic story and, more subsequently, guess what humans will find engaging in the story.
AI to Predict Risk of Developing Psychosis
IBM's Computational Psychiatry and Neuroimaging research team is using machine learning to predict the risk of developing psychosis. The team used AI to analyze the speech patterns of 59 individuals who had participated in a study. Transcripts of their interview were broken down into parts of speech and were scored on how coherent the sentences were. A machine learning model developed by them then determines, based on those speech patterns, who is at risk of developing psychosis. The algorithm was also able to differentiate speech patterns of patients who had recently developed psychosis from those of healthy patients with 72 percent accuracy. The program found that those at risk of psychosis used fewer possessive pronouns when speaking and constructed less coherent sentences.
Artificial Intelligence to Develop Artificial Human Brain
Researchers at the U.S. National Institute of Standards and Technology (NIST) have built a superconducting switch that "learns" like our biological brain. It learns by processing the electrical signals it receives and producing appropriate output signals. The switch, called a synapse, mirrors the function of biological synapses in the brain. A synapse is a connection between two brain cells and allows neurons to communicate with each other. The artificial synapse is designed to learn even from surrounding environment. Its performance is efficient that its biological counterpart. It uses less energy than our brains do and firing signals much faster than human neurons. This affects processing significantly because greater the frequency of electric signals that are fired and received, stronger the connection between two synapse becomes. The switch is will boost the ability of the “neuromorphic computers” which improve decision-making abilities of smart devices such as self-driving cars and even cancers diagnostic tools.
Deep Learning Pioneered for Real-Time Gravitational Wave Discovery
Scientists at the National Center for Supercomputing Applications (NCSA) have pioneered the use of GPU-accelerated deep learning for rapid detection and characterization of gravitational waves. It will allow astronomers to study gravitational waves using minimal computational resources, thus saving resources and time. Researchers Daniel George and Eliu Huerta produced “Deep Filtering”, an end-to-end time-series signal processing method. They combined deep learning algorithms with numerical relativity simulations of black hole mergers and the data from the LIGO Open Science Center. In comparison to available gravitational wave detection algorithms, Deep Filtering gives similar results with lower errors.