Artificial intelligence helps create at the right time
Ana Manasovska is working on improving the semantic recognition and recommendation platform for the inventors. © 2018 Alain Herzog
Summer series. Student project (9/9). By using artificial intelligence to comb through the vast array of published research and detect the findings most relevant for invention, engineers can magnify their creative ability and invent faster and more disruptively than has been previously possible. This is the approach that Ana Manasovska helped develop as a Master’s student at EPFL, and the one used by creative Artificial Intelligence firm Iprova, based at EPFL’s Innovation Park, to come up with a wide range of inventions. Manasovska, whose Master’s research involved testing different phrase recognition methods, now works for the firm.
Inventions like sensors for self-driving cars that can monitor passengers’ health, a geolocation system that can help smooth out passenger traffic on public transportation, and a smartphone feature for virtually painting the light ambience of a room involve pulling together data from several research fields in an inventive way. The ever increasing amount of information in the world, spread across many different industries and markets, makes this an increasingly difficult task. To make it possible for inventors to sense the inventive signal in this ever increasing amount of noise, AI researchers and software developers at Iprova – the innovation creation firm that came up with the aforementioned inventions – have developed an artificial intelligence platform that includes sophisticated semantic analysis algorithms. Ana Manasovska helped create this program as part of her Master’s degree in computer science at EPFL, in association with the school’s Artificial Intelligence Laboratory. She now works for the company, which is based at EPFL’s Innovation Park, to further develop the software that makes it easier for engineers to invent faster and more disruptively..
Millions of publications sifted
Millions of research, industry news and other articles are published around the world every year. One part of Iprova’s artificial intelligence platform works by performing a semantic analysis of the terms in published articles. Manasovska’s thesis on summarization methods contributed to this by testing various phrase recognition methods, which she did by representing individual phrases as vectors. If two phrases have a close virtual spatial location, then their meanings are probably similar. This technique can be used to generate better summaries by measuring phrases’ semantic similarity.
What Manasovska found when comparing the different methods is that the more complicated architectures weren’t necessarily better suited to this task. “Even with a simple architecture, we got excellent results in identifying phrases with similar meanings,” she says. “We also learned that the best way to generate the kinds of summaries that Iprova needs is to approach them from an inventor’s perspective. Most conventional summary-generation methods don’t do that, which is why we wanted to develop our own,” she adds.
Today Manasovska is working on further improving the semantic recognition and recommendation platform. She works closely with inventors, aiming to find out what kind of data they need and how they plan to use it. She has developed programs allowing engineers to create inventions using input data that they wouldn’t have been able to easily get otherwise. An example of this is the linking of information from inventively relevant, but otherwise disparate, research fields that open up entirely new invention opportunities. “What I really like about my work is that it lets me stay on top of the latest developments in machine learning and natural language processing (NLP) – two fields that are advancing rapidly. I have the opportunity to use the latest technology and the power of data to help people spot relevant new findings more efficiently,” she says.
“Traditional inventors were scientists or engineers with a deep understanding of a specific technical field. This only gave the inventor access to a limited amount of research insight. Even collaborative inventing through teamwork only provides insight into a handful of additional fields, since it’s just a team of specialists. With such approaches to invention, researchers can only dig deeper into specific areas rather than offering genuine innovation by taking the field in a different direction.
Iprova does this on a massive scale – in real-time – by using data from across the spectrum of human knowledge to make connections between ideas from different fields of study.”says Julian Nolan, CEO of Iprova. The company is combining AI with a team of creative scientific minds – the invention developers – to accelerate the development of tomorrow’s products and services. Its customers include some of the best known technology companies in Silicon Valley, Japan and Europe. Hundreds of patents have been filed based on its inventions, which are cited by companies including Google, Microsoft and Amazon.