Brain-Score

Brain-Score 平台旨在产生强大的腹侧流计算模型。我们使研究人员能够快速了解他们的模型在多个维度上与标准化大脑基准的得分情况，并有助于与其他最先进的模型进行比较。同时，新的大脑数据可以快速针对各种模型进行测试，以确定现有模型对数据的解释能力。 Brain-Score 由 Brain-Score 团队与世界各地的研究人员和实验室合作组织。我们正在努力打造一个易于使用的平台，在该平台上，可以轻松提交模型以在一系列大脑基准测试中获得分数，并且可以合并新的基准来挑战模型。这种量化的方法让我们可以使用不同的评估技术（指标）在一系列实验（数据）中跟踪我们的模型与大脑的接近程度。更多详细信息，请参阅技术论文和观点论文。The Brain-Score platform aims to develop robust computational models of the ventral visual stream. We enable researchers to quickly obtain the scores of their models against standardized brain benchmarks across multiple dimensions, and facilitate comparisons with other state-of-the-art models. Meanwhile, new brain data can be rapidly tested against various models to evaluate the explanatory power of existing models for these data. Brain-Score is organized by the Brain-Score team in collaboration with researchers and laboratories across the globe. We are striving to build an easy-to-use platform where models can be conveniently submitted to acquire scores across a suite of brain benchmark tests, and new benchmarks can be integrated to challenge these models. This quantitative approach allows us to track how closely our models align with the brain across a range of experiments (datasets) using different evaluation techniques (metrics). For more details, please refer to the technical and perspective papers.

搜集汇总

背景与挑战

背景概述

Brain-Score是一个旨在生成强大腹侧流计算模型的开源平台，允许研究人员评估模型在标准化大脑基准上的得分并与先进模型比较。该数据集由多个研究机构于2018年发布，专注于量化模型与大脑的接近程度。