Thought Leadership
Patents
- Patent pending (filed & accepted) – Selective data transfer utilizing network feature representation learning
AI Research
Published Articles
Active Research
-
Data Driven Modeling for Structural Reliability
Application of machine learning to predict low cycle fatigue of aerospace components
- Low cycle fatigue life is estimated using elastic or plastic finite element stress analysis results for an aerospace part during the design phase. This takes into account the operating temperatures, minimum and maximum loading conditions, high cycle fatigue interaction and geometric characteristics. During scheduled or unscheduled part maintenance, it is necessary to reassess remaining useful LCF life. This poses an operational challenge that often takes over a day to calculate a single sample prediction requiring specialist input. To address this challenge, machine learning can be employed for LCF life prediction in a matter of seconds under known design and operating conditions
-
Machine Learning for Telecommunication Networks
Machine Learning for Cable Telecommunications to Enhance Customer Experience
- From work order entry to ensuring uninterrupted service, network operators manage complex networks. Substantial growth in high speed personal devices are not only driving huge bandwidth demands, but also resources of various data centers connected to these networks. Operator’s digital transformation journey is a move from manual to automated processes. One of the key factors for success will depend on how well they can transform the customer experience. This is where machine learning can help operators better understand and serve their customers. This research explores the use of machine learning methods to deliver enhanced customer experience for cable operators. It examines a use case based approach in building intelligence into each step of the customer and the operator journey. We analyze customer perspectives in the context of use case, such as their need for a simplified and cost-effective experience, their demand for real time response and an intuitive interface, and their willingness to pay for a seamless interaction experience. We then look at the operator’s need for customer acquisition and retention, their demand for reduction in variable costs and increase in operational efficiencies of their business and networks. In addition to identifying use cases that directly impact customer experience, we dive deep into each of its data and machine learning problem frameworks and recommend architectures to help solve these problems.
-
Research in Autism Spectrum Disorder
Using computer vision to track behaviors and aid therapy for young children with autism
- Autism spectrum disorder (ASD) is a developmental disability that can cause significant social, communication and behavioral challenges. Autism affects more than 1% of children and is growing in prevalence at a rate of about 10% per year. Early therapy at a very young age, focused on social skills, has been proven to reduce the most debilitating symptoms of the disorder in later life. These treatment interventions can include behavior therapy, speech-language therapy, play-based therapy and more. However, with any treatment for young children it can be difficult for parents, therapists and physicians to remain objective in tracking progress. Knowing what is working and what isn’t is absolutely critical for fine tuning interventions. Much research emphasis has been placed on finding simple, easy to measure metrics but an underlying problem will always remain; tracking metrics regularly, accurately and without bias is difficult for human beings. Given that treatments typically last years we feel that a computer vision approach of using artificial intelligence to automate tracking would be hugely beneficial. The goal of this project will be to create a proof of concept (POC) showing that useful metrics can be tracked reliably using the OAK-D AI camera during simulated Applied Behavior Analysis (ABA) therapy.