-
find an example with a more obviously "optimistic" generalization error
- don't use kNN - it's too stable, has only one tuning param
- use more high dimensional / more noisy data (or a smaller tr…
-
## Arxiv/Blog/Paper Link
Paper: https://arxiv.org/abs/2308.13280
Code: https://github.com/clessig/atmorep
## Detailed Description
Self-supervised representation learning that uses historical r…
-
## What problem does this solve or what need does it fill?
Building images from programmatically controlled data can be a useful tool for getting reasonably high performance without having to swap …
-
Online Distribution Shift Detection via Recency Prediction. (arXiv:2211.09916v2 [cs.RO] UPDATED)
https://ift.tt/C3NtJeE
When deploying modern machine learning-enabled robotic systems in high-stakes ap…
-
Hi, I work with high-dimensional data (hundreds of dimensions), and it requires using cosine distance. However, TextureTsne has Euclidean distance [hardcoded](https://github.com/biovault/nptsne/blob/m…
-
Hi,i'm Using Caps-Net to Handle High-dimensional and Small-sample Problems, but train cannot converge。My data dimension is 6670, 2 classes. what should i do ,thanks a lot
-
Can one use the warpSet or gaussNorm function of flowStats to normalize high-dimensional CyTOF data? I have such data with a 36-marker panel for 70 samples and I want to reduce batch effects among the…
-
Terminology: An *embedded set* is a high-dimensional set that is universal in some dimensions. Let `U(m)` be the `m`-dimensional universal set. Then I am talking about a set `U(k) × L × U(l)` with `L`…
-
Especially `plot_highdim` needs refactoring:
- Structure of several completely different types of plots in one function is a bad idea, even if it allows some code sharing. Hard to see at a glance whi…
-
I'd love to read more about how this tool is being used, but I'm blocked from accessing the paper describing it:
Colby, S.M., Chang, C.H., Bade, J.L., Nunez, J.R., Blumer, M.R., Orton, D.J., Bloods…