Barkley AI — Research & Publications https://getbarkley.com/ Barkley is the first behavioral intelligence research platform built around individual baselines rather than population averages. Publications, benchmarks, datasets, and open-source releases. en labs@getbarkley.com (Elodie Aishwarya P. Remoissenet) Fri, 03 Jul 2026 00:00:00 GMT Head-to-head Validation v2.0 — Individual Baseline vs Breed Average https://doi.org/10.5281/zenodo.20754351 https://doi.org/10.5281/zenodo.20754351 Elodie Aishwarya P. Remoissenet Reference-class benchmark: same detector, two reference frames. AUC 0.988 vs 0.935, declines caught 100% vs 81%, ~34-day median lead time, reproduced across 30 seeds. Synthetic data, DOI archived. Barkley Reference Architecture — the temporal + individual-baseline stack https://github.com/labs-barkley/barkley-reference-architecture https://github.com/labs-barkley/barkley-reference-architecture Elodie Aishwarya P. Remoissenet Open reference architecture for behavioral-drift detection: individual baselines, temporal modeling, drift engine, silence layer, reference class, behavioral replay, head-to-head validation, synthetic validation. Synthetic DogGraph Sample — open longitudinal canine behavioral dataset https://huggingface.co/datasets/labs-barkley/synthetic-doggraph-sample https://huggingface.co/datasets/labs-barkley/synthetic-doggraph-sample Elodie Aishwarya P. Remoissenet Synthetic longitudinal canine behavioral dataset (activity, sleep, social, nocturnal signals over time) for individual-baseline and behavioral-drift research. CC BY-NC 4.0. DogGraph Demo — schema-constrained GraphRAG over a behavioral memory layer https://doggraph.getbarkley.com/ https://doggraph.getbarkley.com/ Elodie Aishwarya P. Remoissenet A graph-native view of the behavioral memory layer: dog → baseline → context → drift → route → compatibility. Read-only Cypher traversal on synthetic DogGraph data. Your Model Doesn't Have a Bias Problem. It Has a Reference Class Problem. https://datadriveninvestor.com/articles/your-model-doesn-t-have-a-bias-problem-it-has-a-reference-class-problem https://datadriveninvestor.com/articles/your-model-doesn-t-have-a-bias-problem-it-has-a-reference-class-problem Elodie Aishwarya P. Remoissenet Featured analysis on DataDrivenInvestor: the reference class is the hidden variable of machine learning. Your Dog Can Be Normal for Its Breed and Abnormal for Itself https://medium.com/@labs-barkley/your-dog-can-be-normal-for-its-breed-and-abnormal-for-itself-0a4c9a7b3f58 https://medium.com/@labs-barkley/your-dog-can-be-normal-for-its-breed-and-abnormal-for-itself-0a4c9a7b3f58 Elodie Aishwarya P. Remoissenet Why individual baselines detect the behavioral drift that population statistics are built to miss. The Normative Trap: Temporal Identity and the Failure of Population Intelligence https://books2read.com/normative-trap https://books2read.com/normative-trap Elodie Aishwarya P. Remoissenet Tue, 02 Jun 2026 00:00:00 GMT Founder's manifesto on temporal identity, behavioral drift, and the failure of population-based AI to understand the living individual. DOI 10.5281/zenodo.20516821. Framework Papers — 11 DOI-archived working papers https://orcid.org/0009-0004-6031-659X https://orcid.org/0009-0004-6031-659X Elodie Aishwarya P. Remoissenet Eleven Zenodo-archived framework papers on behavioral intelligence, individual baselines, longitudinal modeling, and reference classes.