Abstract

Artificial Intelligence (AI) has been investigated since the 1940s, before there were electronic computers and became a research field in 1956 at a workshop at Dartmouth where the term “Artificial Intelligence” was first used. There have been many disappointments in the quest to build AI systems; however, the last 10 years, extraordinary capabilities have been demonstrated. The building blocks for AI are High-Performance Computers and Big Data which are used to estimate parameters of, or train, algorithms called Artificial Neural Networks (ANNs) to execute AI related functions. The amount of data required to train an ANN increases as the number of parameters increases. Big Data allows ANNs to have many layers, so they are called Deep Networks and Deep Learning refers to training Deep Networks. Deep Networks and Deep Learning have become overly hyped with many who extol their virtues and few that describe the weaknesses. Unfortunately, they can be unstable and produce bizarre outputs. This talk covers inherent and difficult flaws in current AI systems and some methods for mitigating the flaws. Examples from environmental sensing will be given.

Meet the Presenter

Paul Gader received his Ph.D. in Math in 1986. He is professor and former Chair of Computer and Information Science and Engineering and an affiliate Professor in Environmental Engineering at the University of Florida.  He has been working on AI for multi-dimensional signal and image analysis since 1984. Dr. Gader has processed data from many sensors, including Radar, Acoustic-Seismic and Underwater Acoustic, Thermal, and Multi/Hyperspectral sensors. He chaired the IEEE Workshop on Hyperspectral Image and Signal Processing (WHISPERS) in 2013 and gave tutorials on sub-pixel analysis at WHISPERS in 2016 & 2018.  He has worked with neural networks since 1990.  Dr. Gader has published many papers and was named a Fellow of the Institute of Electrical and Electronic Engineers (IEEE) in 2011.

Click here to reserve your spot.