Radiocarbon dating of archaeological carbonates from seven cultural stages of Dholavira, Great Rann of Kachchh (GRK), the largest excavated Harappan settlement in India, suggests the beginning of occupation at ~5500 years BP (pre‐Harappan), and continuation until ~3800 years BP (early part of the Late Harappan period). The settlement rapidly expanded under favourable monsoonal climate conditions when architectural elements such as the Citadel, Bailey, Lower and Middle Town were added between the Early and mid‐Mature Harappan periods. Abundant local mangroves grew around the GRK sustaining prolific populations of the edible gastropod Terebralia palustris. Oxygen isotope (δ18O) sclerochronology of Early Harappan gastropod shell suggests seasonal mixing of some depleted (δ18O ~ −12‰) river water in summer/monsoon months (through ancient Saraswati and/or Indus distributary channels) with seawater that periodically inundated the GRK. Evaporation from this semi‐enclosed water body during the non‐monsoon months enriched the δ18O of water/shell carbonates. The humid fluvial landscape possibly changed due to a catastrophic drought driving the final collapse of the settlement of Dholavira exactly at the onset of the Meghalayan (Late Holocene) stage (~4300–4100 years BP). Indeed, Dholavira presents a classic case for understanding how climate change can increase future drought risk as predicted by the IPCC working group. Copyright © 2019 John Wiley & Sons, Ltd.

We propose a novel high-performance, parameter and computationally efficient deep learning architecture for tabular data, Gated Additive Tree Ensemble(GATE). GATE uses a gating mechanism, inspired from GRU, as a feature representation learning unit with an in-built feature selection mechanism. We combine it with an ensemble of differentiable, non-linear decision trees, re-weighted with simple self-attention to predict our desired output. We demonstrate that GATE is a competitive alternative to SOTA approaches like GBDTs, NODE, FT Transformers, etc. by experiments on several public datasets (both classification and regression).

No abstract

No abstract

The Minoan eruption of Santorini, Greece, is an important and often-debated chronological marker in contexts of the Eastern Mediterranean region. Among various age estimates of this event, one based on wiggle-matching of radiocarbon (14C) dates from an olive branch found in Santorini by Friedrich et al. (2006) has been widely discussed. Calibrated age estimates based on wiggle-matching of these 14C ages have been changing with improvements in the 14C calibration curve. As also shown earlier, calibration of average 14C age of multiple tree rings dated together should not be done using a single-year calibration curve. Since recent calibration curves include many single-year 14C datasets, a different approach should be considered to calibrate the average 14C age of block of multiple tree rings. Here we have demonstrated the use of multiple moving average (MA) calibration curves for calibrating the sequence of four 14C ages reported for the Santorini olive branch. The resultant calibrated ages for the Minoan Eruption are relatively younger than previous estimates and range from the late-17th century BCE to mid-16th century BCE date.