Fast Approximation of Similarity Graphs with Kernel Density Estimation
Authors: Peter Macgregor, He Sun
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we empirically evaluate the performance of spectral clustering with our new algorithm for constructing similarity graphs. We compare our algorithm with the algorithms for similarity graph construction provided by the scikit-learn library [20] and the FAISS library [11] which are commonly used machine learning libraries for Python. |
| Researcher Affiliation | Academia | Peter Macgregor School of Informatics University of Edinburgh United Kingdom, He Sun School of Informatics University of Edinburgh United Kingdom |
| Pseudocode | Yes | Algorithm 1 SAMPLE, Algorithm 2 FASTSIMILARITYGRAPH |
| Open Source Code | Yes | The code to reproduce our results is available at https://github.com/pmacg/kde-similarity-graph. |
| Open Datasets | Yes | Finally we study the application of spectral clustering for image segmentation on the BSDS dataset. The dataset contains 500 images with several ground-truth segmentations for each image. [2] Pablo Arbelaez, Michael Maire, Charless Fowlkes, and Jitendra Malik. Contour detection and hierarchical image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 33(5):898 916, 2011. |
| Dataset Splits | No | The paper discusses the datasets used and some parameters for algorithms but does not explicitly provide details about training, validation, or test dataset splits. |
| Hardware Specification | Yes | All experiments are performed on an HP ZBook laptop with an 11th Gen Intel(R) Core(TM) i7-11800H @ 2.30GHz processor and 32 GB RAM. |
| Software Dependencies | No | The paper mentions software like C++, Sci Py [30], and stag [17] libraries, but it does not provide specific version numbers for these software dependencies. |
| Experiment Setup | Yes | For each experiment, we set k = 10 for the approximate nearest neighbour algorithms. For the synthetic datasets, we set the σ value of the Gaussian kernel used by SKLEARN GK and OUR ALGORITHM to be 0.1, and for the BSDS experiment we set σ = 0.2. |