A team o scientists fae the University o Glesga his unveiled a grundbrakin saftware model aat promises tae mak e analysis o geospatial data mair eco-freendly. E model, cryed ‘GeoAggregator’, eeses machine lairnin for tae lessen e computin pouer needed tae haandle vast an intricate geospatial datasets.

As GPS an satellite data become mair common, e warld gaithers massive amoonts o geospatial information ilka day. Yet, makkin sense o this data is nae sma feat. Traditional methods an even some AI models struggle tae grasp e complex spatial ties ithin e data. GeoAggregator tackles thae hurdles wi a lichtwecht AI model aat deftly examines foo nearby places affect ane anither an foo paitterns shift fae ae spot tae anither. It is faster, mair scalable, an demands fewer resoorces than e eens afore it, appenin doors for researchers an policymakkers alike.

Ae staundout feature o GeoAggregator is its Gaussian-biased local attention mechanism. Iss allous e model tae zero in on nearby data pynts fyle haadin e bigger picture in mind. Sic an approach boosts predictions for various spatial data issues, fae forecastin air pollution tae spottin hoosin price trends an studyin poverty spreid.

E model’s Cartesian attention mechanism keeps it licht fyle ensurin heich accuracy. Even as datasets grow, GeoAggregator processes them wi ease, a feat mony traditional AI models cannae match. Tests shaw aat GeoAggregator performs as weel as or better than ither leadin methods, eesin far fewer resoorces. Iss maks it a practical chyce for real-warld tasks far speed an efficiency are key.

E researchers hae shared their code as appen-soorce, urgins ither fowk tae eese an refine it. As interest in location-based data swells, tools like GeoAggregator can turn complex geospatial information intae meaninfu insichts aboot foo fowk an places interact.

Dr. Mingshu Wang fae the University o Glesga’s School o Geographical & Earth Sciences, a co-author o the study, pynts oot e model’s potential: “GeoAggregator merks a muckle lowp in makkin complex data analysis mair efficient an accessible. We did aa oor data analysis on a single laptop, shawin juist foo effective iss tool is.”

E team is warkin on an appen-soorce Python package tae mak e model freely available. Iss effort is pairt o ongaan PhD research by Rui Deng, fa will continue tae enhance the model ower e neist twa years. Dr. Ziqi Li, an Assistant Professor at Florida State University, is a co-author o the paper.

This research is pairt o the University o Glesga’s growin expertise in AI an machine lairnin, follaein e laanch o their Centre for Data Science & AI. E School o Geographical & Earth Sciences his recently received fundin tae lead e Exascale computin for Earth, Environmental, an Sustainability Solutions consortium, fit will train new PhD students tae develop saftware for environmental applications on neist-gen computin systems. E team’s paper will be presented at e AAAI Conference on Artificial Intelligence iss wikk.