Practical applications of deep learning algorithms enhances the fields of archaeology and history.
Archaeologists are identifying new archaeological sites and other potential locations of interest using deep learning techniques. Traditionally, a human would pour over data and make the determination for the location of a site. Deep learning makes this process much faster.
SEE: IT leader's guide to deep learning (Tech Pro Research)
With the wide use of remote sensor data in archaeology, deep learning allows the archaeologist to use the machine for much of the grunt work humans previously did. With the image recognition abilities of machine learning, this means more time can be spent on verification than on identification.
"I found that the way that it's been done before could probably be improved with the use of deep learning," said Iris Kramer, a computer science Ph.D. student at the University of Southampton in the United Kingdom.
Kramer, who earned her bachelor's and master's degrees in archaeology, saw the potential for deep learning algorithms in the field of archaeology because of its ability for image recognition.
"My interest in deep learning came at the end of my master's really, when I did my dissertation about automated detection of archeology on remote sensing data," Kramer said. "I saw deep learning emerging and if self-driving cars could be on the road, then why couldn't we maybe do this with archeology?" That interest led Kramer to pursue a Ph.D. program that combined machine learning with archaeology, teaching herself to code the year between earning her masters and Ph.D.
Now, Kramer uses deep learning algorithms with remote sensor datasets in order to enhance how archaeologists work with LiDAR, a 3D laser scanning technology. LiDAR measures the distance to a target with illuminating pulsed laser lights, and then measures the reflected pulses with a sensor. It is combined with established archeological data to detect patterns, which can indicate potential archaeological points of interest.
"We have new detections of possible sites, but also we can now, rather than just search for one object and make a whole group based approach for this one object; we can now easily switch to a new object if we want to detect a new type of site that we're interested in," Kramer said.
"It's like a tool for desk-based assessment," Kramer continued. "If they [archaeologists] are interested in certain objects; if there is enough of those sites, then this will just be another way to find new sites, and that is just for remote sensor data."
Kramer adds that deep learning and the abilities we see with it now works on small datasets as well.
Other fields within archeology can benefit from deep learning as an integrated tool, along with other approaches. "There is obviously a lot of archeological sites that have not been discovered yet, and with techniques like these, it almost becomes way more interesting to find a quicker way to do things," Kramer said.