ALSH for solving mips sublinearly
################
DATASET
################
Link: http://files.grouplens.org/datasets/movielens/ml-latest-small.zip
To execute the code on a small dataset :
- run ./dataset_small.sh
- in ipynb files comment dataset = "datasets"+os.path.sep+"ml-latest" and uncomment dataset = "datasets"+os.path.sep+"ml-latest-small"
To execute the code on full dataset :
- run ./dataset.sh
- in ipynb files uncomment dataset = "datasets"+os.path.sep+"ml-latest" and comment dataset = "datasets"+os.path.sep+"ml-latest-small"
################
PAPER
################
Link to the paper : http://papers.nips.cc/paper/5329-asymmetric-lsh-alsh-for-sublinear-time-maximum-inner-product-search-mips.pdf
#################
CODE
#################
ALSH.mips has stand alone code for complete implementation of LSH, SLSH and ALSH (helper file: utils.py)
#####
Files:
#####
-
plots_paramas2.ipynb has functions for determining best parameters for U, m, r and c
It also has plots that show minimisations of rho functions. -
text.py has code to assemble netflix data from the zip file(s)
-
utils.py has code for helper functions of L2LSH, SLSH and ALSH
-
ALSH_MIPS.ipynb has code that tests on L2LSH, SLSH and ALSH.
It also has precision recall curves