MINERAL EXPORATION — Neural networks applied to Carlin trend

By applying neural pattern recognition technology, Uranerz USA is reinterpreting the Ren gold property in the Carlin trend of northern Nevada.

The property is a joint venture between Uranerz, Romarco Minerals (R-T) and Homestake Mining (HM-N).

The reinterpretation of what has been considered a typical Carlin-type sediment-hosted gold deposit is the result of initial testing of a neural pattern recognition program developed by US Mineral Labs (USML) of Auburn, Calif.

The program, called “X-net,” was developed by William Henderson, chairman of USML, after the company experienced problems in geochemical modelling. In mineral deposits, the relationships between elements vary over distance and depth, making correlations difficult. X-net attempts to solve the problem by means of so-called “neural networks,” which are uniquely suited to recognize complex patterns in large data packages.

“It is a way to look inside a set of data for the not-so-obvious patterns,” Henderson says.

Neural networks are a form of artificial intelligence and are currently being applied in the medical field to screen for cancer and other diseases. Neural pattern recognition programs are also common in the financial world and in the optical character recognition and speach recognition programs.

In the case of mining, Henderson says the technology has proved effective in stepout drilling from an existing mine.

At the Ren property, Uranerz has put Henderson’s program through a test case.

Geologists took some 700 soil samples over a 1-sq.-mile area. The data were put into the program, which examined the information across 52 discreet variables, analyzing the complex relationships between them. The resulting map

showed a concentric pattern of anomalies.

Kelly Cluer, a geologist for Uranerz, says that while the map did not define any drill targets, the results gave him confidence in the program’s ability to assemble a picture of the deposit’s alteration system. The neural network data compared favorably to data compiled from previous geophysics and structural mapping in the area. “The results from the network were geologically meaningful,” Cluer asserts.

The concentrically zoned pattern of alteration at the Ren property appears to indicate a possible gold porphyry at depth.

Cluer says the use of neural pattern recognition could radically change the way we look at deposits in the Carlin trend. Henderson says X-net is also being applied successfully to properties in Nevada’s Battle Mountain trend.