Understanding the Irish Famine Using Deep Neural Networks and Protolanguage

Author(s): Shaiyon Merkel

Year: 2017


Drawing from historical records and archaeological data, we used multilayer neural networks to construct a sociocultural model of the Irish Famine.  We found that Capital Exchange optimization for non-elites frequently contained polynomial-time mappings to the Assignment and Knapsack problems (which are both NP-hard).  However, we only occasionally encountered nontrivial instances of these mappings when the same algorithms were applied to elites.  That pattern of asymmetric computational complexity was reproducible even when resource loss was reduced to survivable levels.  This indicates that the Irish Famine may not have been attributable entirely to insufficient food resources; information advantages encoded within social structures may also have been a contributing factor.  We applied our findings to develop the experimental protolanguage model of computation, with which we successfully averted the emergence of starvation conditions in our model.  While protolanguage computing cannot help those who suffered during the Famine, it can be applied to predict and avert future tragedies.

Cite this Record

Understanding the Irish Famine Using Deep Neural Networks and Protolanguage. Shaiyon Merkel. Presented at Society for Historical Archaeology, Fort Worth, TX. 2017 ( tDAR id: 435487)

This Resource is Part of the Following Collections

Spatial Coverage

min long: -129.199; min lat: 24.495 ; max long: -66.973; max lat: 49.359 ;

Individual & Institutional Roles

Contact(s): Society for Historical Archaeology

Record Identifiers

PaperId(s): 566