 Our fifth presenter is George Lederman, whose title is Good Vibrations, How to Protect Our Infrastructure. In 2012, a first-grade New Jersey collapse, sending both trains and thousands of gallons of chemicals into a sensitive wetland area. Conrad, the British owner, knew the 72-year-old bridge was deficient, but they didn't have the kind of objective information to make the difficult decision to close the bridge entirely. In fact, there are tens of thousands of structures like this throughout our infrastructure that managers have to struggle with every day deciding whether to close them or keep them open. My research focuses on economical ways that we could monitor our entire infrastructure continuously providing objective information to planners. Our system works like this. We've been placing sensors in order to test our system on Pittsburgh's light rail system. We first acquire data from the vehicle with accelerometers. We're looking at the vibrations of the vehicle. Next, we process it to handle some of the environmental variability. There's significant problems in the infrastructure such as cold temperature, high winds or flooding. We look for features which are robust to this variability. Next, we look at changes. Previous people who have done this kind of work, they have tried to model the way the infrastructure responds. But instead of doing this, we just look for statistical anomalies relative to previous behavior. Finally, because we have this objective information, we're able to provide it in real time to people like infrastructure managers and in order that they can make the best decisions at the moment. These guys in turn can then tell us what information is the most useful to them and we can perform some sort of reinforcement learning. This is a way of tackling the three major challenges which have plagued people in the past. The large size of the infrastructure. These are assets which stand in thousands of kilometers. The environmental variability. And also the uniqueness of civil assets. Unlike other things that we monitor, like aeroplanes or manufacturing equipment, each civil asset is different. So over the last two years, we've been monitoring the Pittsburgh infrastructure from a train. And here's some data that we've been able to find. This is a very low cost element, so any information that we find is useful. Every now and then, the Port Authority adjusts the balance under the tracks, the stones and the tracks wide. And when they performed this last year, we were able to see it. So on the left, we see a 10-kilometer section of track. This is the amount of vibrations in the train and the travels of the infrastructure. Before and after this maintenance procedure has been performed. And if we look at the correlation between the two signals, we can actually see changes after the location where the maintenance work was performed. This is amazing. This is a pretty subtle kind of thing that we're looking for and we're able to see it just from the vibrations in the vehicle. So I'd like to leave you with one last note. Our infrastructure continues to deteriorate each day. But sensors and computational devices like smartphones are becoming more and more basic. If we can harness the information in these devices, we can tell the state of the infrastructure just from the vibrations you feel on the way to work. And we can ensure that classes like this don't happen in the future. Thank you.