Imogen Low, is the Co-Founder & chief technical officer (CTO) at NWO.ai, a predictive platform that tracks more than 20 million microtrends, and notifies clients about trends before they become exponential.
What initially got you interested in Machine Learning?
I first become interested in machine learning when I was in high school. In my spare time, I was developing interactive 3D maps of complex, multi-level campus facilities such as shopping malls, airports, hospitals and schools. These maps could be used to aid in indoor navigation and wayfinding for visitors. I was also interested in collecting behavioral data points about individual users as they move throughout buildings. This information could be used to anticipate the flow of foot traffic over time and optimize navigation routes.
But it wasn’t until I was 17yrs old, when I landed my first job as a machine learning engineer at SAP, that I really became excited about AI. While I was working at SAP, I worked on a range of different problems. For example I analysed satellite data from NASA to predict the quality of crop yield; I used computer vision technology to analyzing drone footage to monitor livestock; I used enterprise transactional data to predict tax fraud; and I combined computer vision technology with natural language processing to build out conversational and augmented reality virtual assistants.
You took an odd path from Australia, to Singapore and finally in America, could you discuss this journey?
The indoor navigation application I developed in high school took me to Taiwan where I showcased it at the Asia Pacific ICT Alliance (APICTA) awards conference. The experience gave me a new perspective on the pace of technological change and innovation happening in Asia — particularly their adoption of artificial intelligence technologies. I wanted to immerse myself in an Asian country and be a part of those advancements. So I took up a machine learning position in the SAP APJ head office in Singapore.
While I was working in Singapore, I developed a virtual assistant that became a leading example of innovation at SAP. So I was invited to showcase it at an innovation summit across Asia. At the summit, I listened to a fascinating keynote presentation from Pulkit Jaiswal, a successful drone entrepreneur. He explained how he could predict geopolitical tension between countries by analyzing online data. I was so intrigued by the idea that I decided to quit my job that day and partner with him as a co-founder in his latest venture, NWO. While the company is based out of New York, I’m based in Sydney.
Could you describe how NWO.ai helps identify trends before they become exponential?
At NWO.ai, we model how the adoption of different ideas spread throughout global networks. The aim is to predict significant upticks in mindshare. To achieve this, we designed a cross-correlation engine that pinpoints which data sources and consumer voices are the strongest leading predictors of the adoption of a trend. We can also identify the main factors influencing growth. We then leverage our dynamic weighting system to amplify the most predictive datapoint and dampen the lagging ones.
What are some of the machine learning technologies that are used?
We leverage a set of state-of-the-art natural language processing (NLP) machine learning models such as document encoders, named entity recognition (NER) models, and machine translation. We combine these with techniques we’ve adopted from association rule mining and graph theory to model our unstructured textual datasets into a time-aware, multi-lingual, semantic knowledge graph. We can then represent how strongly associated different ideas are at different points in time, and dynamically build associations between new data points in real-time.
What type of public or private data sources are used?
We primarily rely on unstructured, publicly available data sources to build our prediction models. These include social media posts, news articles, blogs, forums, comments and search histories. We also combine these with proprietary enterprise transactional data, such as historical sales, demand and inventory data.
What advice do you have for female entrepreneurs?
Now is a great time to be a female founder! Don’t be deterred by the idea that entrepreneurship is predominantly male-dominated. There’s been a greater call for gender diversity in entrepreneurship recently. Women have a lot to offer in terms of competing perspectives on the world and different ways of thinking about solving problems. Because of this, we’re seeing new support networks that encourage women to start businesses and offer guidance, mentorship and investment opportunities.
Thank you for the great interview, readers who wish to learn more should visit NWO.ai.