https://github.com/dnbaker/dci
Dynamic Continuous Indexing
https://github.com/dnbaker/dci
dci indexing lsh pdci
Last synced: about 1 month ago
JSON representation
Dynamic Continuous Indexing
- Host: GitHub
- URL: https://github.com/dnbaker/dci
- Owner: dnbaker
- License: mit
- Created: 2020-04-02T17:50:30.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2020-04-04T13:52:21.000Z (about 6 years ago)
- Last Synced: 2025-03-01T14:33:16.417Z (over 1 year ago)
- Topics: dci, indexing, lsh, pdci
- Language: C++
- Size: 31.3 KB
- Stars: 2
- Watchers: 3
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
### DCI: Dynamic Continuous Indexing
Fast, generic implementations of [Dynamic Continuous Indexing](https://arxiv.org/abs/1512.00442) and [Prioritized Dynamic Continuous Indexing](https://arxiv.org/abs/1703.00440) (PDCI).
Additionally, includes matrix hashes for several kinds of dissimilarity measures. These include:
1. L2 distance (both E2LSH and P-stable LSHashing)
2. L1 distance (P-stable)
3. Total Variation Distance (a special case of P-stable L1)
4. Lp distance, 1 < p < 2, using CMS sampling
5. Jensen-Shannon Divergence
6. S2JSD (Jensen-Shannon Metric, the square root of the JSD)
7. Hellinger Distance
These can be used either in a table, such as `dci::hash::LSHTable`, or for DCI.
The [Fast Randomized Projections](https://github.com/dnbaker/frp) project in which this was originally developed also has an
FHTHasher, which computes the projections using the FHT compatible with this.