Welcome to this new AI Lounge. We are happy to present two new interesting presentations.
Poster about the AI lounge (pdf).
Coffee, tea and cookies will be served.
Looking forward to seeing you!
Intelligent geoscanning for cheaper roads: Combining artificial neural networks and airborne geophysics for more efficient infrastructure construction.
Abstarct
Globally, most large infrastructure projects go over budget, typically exceeding their budgeted cost by 20-50%. Much of this can be attributed to unforeseen geological problems. Thorough ground investigations reduce this risk, but geotechnical drillings and laboratory tests are expensive, are time consuming, and give information at only one point. In contrast, airborne electromagnetics (AEM) is a low-cost geophysical method that can quickly cover a large area in a small amount of time. However, extracting engineering parameters from these complex data is challenging. In this lecture, Craig will discuss how machine learning methods have been the key to unlocking the value of AEM data for geotechnical projects. The focus will be on how artificial neural maps have been applied to model depth to bedrock, a key parameter in many projects. Such models rival the accuracy of manual interpretations by experts and are markedly more accurate than existing automated methods. Craig also will discuss both the impact of implementing introducing ANN has had on the role of airborne methods in the geotechnical sector and also the future outlook for using AI to extract geotechnical information from geophysical data.
by Craig W. Christensen (EMerald Geomodelling)
Biography
Craig W. Christensen is VP Technology and Co-Founder of EMerald Geomodelling. Craig's talent is bridging the gaps between seemingly disparate fields within applied geoscience, having worked with various topics within geotechnical engineering, geophysics, hydrogeology, and geostatistics. He started his career at the Norwegian Geotechnical Institute as a summer intern in 2013 and was researcher and consultant from winter 2018 to summer 2019. He completed his MSc in Geology and Geophysics at the University of Calgary (2017). He also holds a BScE in Geological Engineering from Queen’s University (2014).
Big data serving: Processing and inference at scale in real time
Abstract
Applying machine learning online requires solving the problem of model serving: Evaluating machine-learned models over data points in real time when a decision is needed. Tools such as TensorFlow serving can solve this problem when it is sufficient to evaluate a model over a single data point per decision. However, in many use cases it is necessary to evaluate over many data points to make a decision. This is the problem of big data serving - an inherently hard problem for architectural reasons and one not usually addressed by such tools. This talk explains the problem, well-known use cases such as search and recommendation and introduces the vespa.ai platform which provides an open source solution to this problem.
by Jon Bratseth (vespa.ai)
Biography
Jon Bratseth is a distinguished architect in Verizon, and the architect and one of the main contributors to Vespa.ai, the open big data serving engine.Jon has 20 years experience as architect and programmer on large distributed systems. He has a master in computer science from the Norwegian University of Science and Technology
Questions