lFCI Results: Intrinsic Dimension Estimates for Neural Activity Manifolds

5.5 KB of results from the local Full Correlation Integral (lFCI) estimator applied to neural activity data. The dataset, authored by Jacopo Fadanni and last updated in April 2026, contains estimates of intrinsic dimension for synthetic benchmarks, recurrent neural network tasks, and experimental visual response data. It is intended to provide a robust tool for analyzing the latent geometry of high-dimensional neural spaces.

Use Cases

• Benchmarking intrinsic dimension estimators based on synthetic data from Altan et al., 2019
• Analyzing the relationship between task structure and neural manifold dimensionality based on results from the 'cog-Task' battery
• Investigating the dimensionality of visual response manifolds based on the reference experimental dataset by Stringer et al., 2019

Strengths

• Dataset is small (5.5 KB), facilitating quick download and inspection.
• Results are derived from a proposed robust pipeline (lFCI) designed to overcome limitations of common estimation methods.
• Applies to multiple data sources: synthetic benchmarks, trained neural networks, and experimental recordings.

Limitations

• Row count and column-level documentation are unknown; field semantics must be inferred after download.
• The dataset's scope is limited to the specific results of the lFCI method on the described cases.

Provenance

Source

figshare

Collection Method

Results generated by applying the lFCI estimation pipeline to specified datasets.

Time Range

Covers results from referenced studies (e.g., Altan et al., 2019; Stringer et al., 2019).

Freshness

Last updated 2026-04-03 17:45:29

Geography

null