Fast k-Nearest Neighbour Search via Prioritized DCI

Authors: Ke Li, Jitendra Malik

ICML 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We also demonstrate empirically that Prioritized DCI significantly outperforms prior methods. In particular, relative to Locality-Sensitive Hashing (LSH), Prioritized DCI reduces the number of distance evaluations by a factor of 14 to 116 and the memory consumption by a factor of 21.
Researcher Affiliation Academia 1University of California, Berkeley, CA 94720, United States. Correspondence to: Ke Li <ke.li@eecs.berkeley.edu>.
Pseudocode Yes Algorithm 1 Data structure construction procedure
Open Source Code No The paper does not provide any explicit statements about releasing source code or links to a code repository for the described methodology.
Open Datasets Yes Evaluation is performed on two datasets, CIFAR100 (Krizhevsky & Hinton, 2009) and MNIST (Le Cun et al., 1998).
Dataset Splits Yes We evaluate performance of all algorithms using crossvalidation, where we randomly choose ten different splits of query vs. data points. Each split consists of 100 points from the dataset that serve as queries, with the remainder designated as data points.
Hardware Specification No The paper mentions 'wall-clock time' for experiments but does not provide any specific hardware details such as CPU/GPU models or memory specifications used for running the experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes For LSH, we used 24 hashes per table and 100 tables... For product quantization, we used a data-independent codebook with 256 entries... For standard DCI, we used the same hyparameter settings used in (Li & Malik, 2016) (m = 25 and L = 2 on CIFAR-100 and m = 15 and L = 3 on MNIST). For Prioritized DCI, we used two different settings: one that matches the hyperparameter settings of standard DCI, and another that uses less space (m = 10 and L = 2 on both CIFAR-100 and MNIST).