Relevant to the techies here, machine learning in archeology sounds like an interesting niche: > Recently, we developed a spectroscopic “thermometer” based on

Relevant to the techies here, machine learning in archeology sounds like an interesting niche: > Recently, we developed a spectroscopic “thermometer” based on Raman spectroscopy and deep learning (DL) algorithms to estimate the exposure of flint artifacts to heat, independent of visual indications of heat exposure [...] > Temperatures were estimated using a 1D-CNN and a fully connected artificial neural network (FC-ANN), as described elsewhere. When compared to FC-ANN (20), the 1D-CNN model reduces the mean absolute error (MAE) between the true and estimated temperatures on the validation experiment from 118 to 103 °C