Travel News

How archaeologists are using deep learning to dig deeper

Finding the tomb of an ancient king full of golden artifacts, weapons, and elaborate clothing seems to be every archaeologist’s fantasy. But looking for them, Gino Caspari can tell you, is incredibly tedious.

Dr. Caspari, a research archaeologist at the Swiss National Science Foundation, studies the ancient Scythians, a nomadic culture whose warriors on horseback terrorized the plains of Asia 3,000 years ago. The tombs of Scythian royalty contained much of the fabulous wealth that they had plundered from their neighbors. From the time the bodies were interred, these graves were popular targets for thieves; Dr Caspari estimates that over 90 percent of them have been destroyed.

He suspects that thousands of graves are spread across the Eurasian steppes, which span millions of square kilometers. He had spent hours mapping the graves using Google Earth images of territories in what is now Russia, Mongolia, and western China’s Xinjiang Province. “It‘s basically a dumb task,” Dr Caspari said. “And that’s not what a well-trained academic should do.”

In fact, a neighbor of Dr. Caspari at International House in Manhattan’s Morningside Heights neighborhood had a solution. Neighbor, Pablo Crespo, then a graduate student in economics at New York City University who worked with artificial intelligence to estimate commodity price volatility, told Dr Caspari what he needed was a neural network convolutional to find its satellite. images for him. The two bonded over a common academic philosophy, to make their work openly available for the benefit of the wider scientific community, and a love of heavy metal music. Over the beers of the International House bar, they began a collaboration that put them at the forefront of a new type of archaeological analysis.

A convolutional neural network, or CNN, is a type of artificial intelligence designed to analyze information that can be processed in the form of a grid; it is particularly well suited to the analysis of photographs and other images. The network sees an image as a grid of pixels. The CNN that Dr. Crespo designed begins by assigning each pixel a rating according to its degree of red, then another for green and for blue. After evaluating each pixel based on a variety of additional parameters, the array begins to analyze small groups of pixels, and then successively larger ones, looking for matches or near-matches with the data it has been. trained to spot.

Working in their spare time, the two researchers broadcast 1,212 satellite images over the network for months, instructing it to search for circular stone graves and ignore other circular and mess-like objects such as piles of debris. building and irrigation ponds.

At first, they worked with images that spanned around 2,000 square miles. They used three-quarters of the imagery to train the network to understand what a Scythian tomb looks like, correcting the system when a known tomb was missing or highlighted one that did not exist. They used the rest of the images to test the system. The network correctly identified known graves 98% of the time.

Creating the network was straightforward, said Dr Crespo. He wrote it in less than a month using the Python programming language and at no cost, not including the price of the beers. Dr Caspari hopes their creation will give archaeologists a way to find new graves and identify important sites so they can be protected from looters.

Other convolutional neural networks are starting to automate a variety of repetitive tasks that are typically imposed on graduate students. And they open new windows to the past. Some of the jobs these networks inherit include classifying pottery shards, locating wrecks in sonar images, and searching for human bones that are for sale, illegally, on the Internet.

“Netflix uses this kind of technique to show you recommendations,” said Dr. Crespo, now a senior data researcher for Etsy. “Why shouldn’t we use it for something like saving human history?”

Gabriele Gattiglia and Francesca Anichini, both archaeologists from the University of Pisa in Italy, excavate sites from the time of the Roman Empire, which involves analyzing thousands of broken pieces of pottery. In Roman culture, almost all types of containers, including kitchen vessels and amphorae used to ship goods around the Mediterranean, were made of clay, so analysis of pottery is essential for understanding life. Roman.

The task is to compare shards of pottery to pictures in printed catalogs. Dr Gattiglia and Dr Anichini estimate that only 20 percent of their time is spent excavating sites; the rest is devoted to the analysis of pottery, a job for which they are not paid. “We began to dream of a magical tool to recognize pottery on an excavation,” said Dr Gattiglia.

This dream has become the ArchAIDE project, a digital tool that will allow archaeologists to photograph a piece of pottery in the field and have it identified by convolutional neural networks. The project, which received funding from the European Union’s Horizon 2020 research and innovation program, now involves researchers from across Europe, as well as a team of computer scientists from Tel Aviv University in Israel who designed the CNNs.

The project involved scanning many paper catalogs and using them to train a neural network to recognize different types of pottery vessels. A second network has been formed to recognize the profiles of the pottery shards. So far, ArchAIDE can only identify a few specific types of pottery, but as more researchers add their collections to the database, the number of types is expected to increase.

“I dream of a catalog of all types of ceramics,” said Dr Anichini. “I don’t know if it’s possible to end in this life.”

Saving time is one of the biggest benefits of using convolutional neural networks. In marine archeology, navigation time is expensive, and divers cannot spend too much time underwater without risking serious pressure-related injuries. Chris Clark, an engineer at Harvey Mudd College in Claremont, Calif., Solves both problems by using an underwater robot to perform sonar scans of the seabed, then using a convolutional neural network to search for images of wrecks and d ‘other sites. In recent years he has worked with Timmy Gambin, archaeologist at the University of Malta, to excavate the bottom of the Mediterranean Sea around the island of Malta.

Their system got off to a bad start: on one of its first trips, they dropped their robot into a wreck and had to send a diver to retrieve it. Things got better from there. In 2017, the network identified what turned out to be the wreckage of a WWII dive bomber off the coast of Malta. Dr Clark and Dr Gambin are currently working at another site that was identified by the network, but declined to discuss details until the research had been peer reviewed.

Shawn Graham, professor of digital humanities at Carleton University in Ottawa, uses a convolutional neural network called Inception 3.0, designed by Google, to search the Internet for images related to the buying and selling of human bones. The United States and many other countries have laws requiring that human bones held in museum collections be returned to their descendants. But there are also bones held by people who circumvented these laws. Dr Graham said he has even seen videos online of people digging graves to fuel this market.

“These people who are bought and sold have never consented to this,” Dr. Graham said. “It continues to violate the communities from which these ancestors were taken. As archaeologists, we should try to stop this.

He made some changes to Inception 3.0 so that it could recognize photographs of human bones. The system had already been trained to recognize objects in millions of photographs, but none of these objects were bones; he has since formed his version on more than 80,000 images of human bones. He’s now working with a group called Countering Crime Online, which uses neural networks to track down images related to the illegal ivory trade and sex trafficking.

Dr Crespo and Dr Caspari said the social sciences and humanities could benefit from integrating information technology tools into their work. Their convolutional neural network was easy to use and freely available for anyone to modify to suit their own research needs. Ultimately, they said, scientific advancements boil down to two things.

“Innovation really happens at the intersections of established fields,” said Dr. Caspari. Dr Crespo added: “Have a beer with your neighbor every now and then.”

[Like the Science Times page on Facebook. | Sign up for the Science Times newsletter.]