feature space
The high-dimensional mathematical space whose axes are the measured variables (e.g. expression of 20,000 genes), within which each sample is a point.
As defined in 251 | Rosemary Braun on Uncovering Patterns in Biological Complexity
Where it comes up · 2
- Tim Cook’s Dashboard2 claims
- 251 | Rosemary Braun on Uncovering Patterns in Biological Complexity2 claims
Most important claims · 4 of 4Search all claims for “feature space” →
- Learning a classifier is equivalent to adjusting the weights of a threshold function so that a hyperplane partitions the feature space between groups; a single neuron is one partition, and the same math scales from two dimensions to 10,000.Tim Cook’s Dashboard
- Deep learning automatically finds a feature space in which data becomes linearly separable, such as mapping an interleaved spiral pair into two separate Gaussian clusters, so that simple partitions can then classify it.Tim Cook’s Dashboard
- Biology has shifted in recent decades from studying organisms and species toward studying systems, networks, and genomes, driven by the revolution in molecular profiling technologies that let researchers assay gene expression in a roughly 20,000-dimensional feature space, creating the challenge of relating those microscopic measurements to macroscopic features seen under a microscope or in a patient.251 | Rosemary Braun on Uncovering Patterns in Biological Complexity
- Although gene expression observations sit in a roughly 20,000-dimensional feature space, those observations lie on a very low-dimensional manifold within it, because not all locations in that space correspond to a system that can live; the goal is to discover the geometry of that manifold and use an observation's location on it to infer the system's behavior.251 | Rosemary Braun on Uncovering Patterns in Biological Complexity