Derek Miller

Title: Distributed Deep Learning on Genomic and Patient Data

Track: Machine Learning/ Analytics

With the recent influx of genomic data due to decreased sequencing costs and greater compute power, researchers are leveraging a wide range machine learning tools to find unique insights from their data. Due to the massive feature space of genomic data, some traditional statistical algorithms are inefficient or struggle to produce meaningful results. At Lifeomic, we have built a highly distributed machine learning system to address the many scalability problems found with processing and analyzing genomic and patient data, leveraging tools such as Apache Spark, Tensorflow, and custom hyper optimization algorithms.

Derek’s first interest in artificial intelligence began at a young age wanting to build a robot that could automatically clean his room, and this ultimately drove him to pursue a professional career in machine learning. Derek’s professional career started as a volunteer developer in Tena, Ecuador developing a health care system for doctors in the Amazon Rain Forest. He later moved back to Indiana and worked at Interactive Intelligence, building NLP engines, customer churn detection services and a large-scale usage analytics system. He now works at Lifeomic, where he primarily works on analytics and machine learning on genomic data.

He continued his academic interests in a professional setting researching the utilization of deep reinforcement learning for NLP. Outside of work he contributes to many open source deep learning frameworks including DeepLearning4j and MxNet. Derek enjoys hiking, playing music, spending time with family, and reading research papers.