Extended Data Table 2 Comparison of the performance of two types of natural language query methods in AIPOM

From: NeuroXiv: AI-powered open databasing and dynamic mining of brain-wide neuron morphometry

  1. To achieve this, we defined three common retrieval scenarios: querying neuron types, querying neurons with particular structures, and querying neurons with specific projection patterns, each comprising 10 test cases. To ensure fair testing, both methods used the same computer configuration, eliminating disparities due to varying computational power. The server-side method involved setting up a local NeuroXiv server on the test computer, while the client-side method accessed the NeuroXiv server hosted on AWS directly. As a result, compared to the server-side approach, the client-side method demonstrated an average response time improvement of 12.3.
Back to article page