PAMI: An Open-Source Python Library for Pattern Mining

Authors: Uday Kiran Rage, Veena Pamalla, Masashi Toyoda, Masaru Kitsuregawa

JMLR 2024 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical This paper tackles this problem by providing a cross-platform open-source Python library called PAttern MIning (PAMI). PAMI provides several algorithms to discover different types of patterns hidden in various types of databases across multiple computing architectures. PAMI also contains algorithms to generate various types of synthetic databases.
Researcher Affiliation Academia R. Uday Kiran EMAIL The University of Aizu Aizu-Wakamatsu, Fukushima, 965-8580, Japan P. Veena EMAIL The University of Aizu Aizu-Wakamatsu, Fukushima, 965-8580, Japan Masashi Toyoda EMAIL Institute of Industrial Science, The University of Tokyo Tokyo, 153-8505, Japan Masaru Kitsuregawa EMAIL Research Organization of Information and Systems, Tokyo, 105-0001, Japan, The University of Tokyo, Tokyo, 113-8654, Japan
Pseudocode No The paper describes the architecture and usage of the PAMI library, providing conceptual explanations and code snippets for implementation, but does not include structured pseudocode or algorithm blocks for the algorithms themselves.
Open Source Code Yes Furthermore, the source code is available under the GNU General Public License, version 3. Finally, PAMI offers several resources, such as a user s guide, a developer s guide, datasets, and a bug report. Keywords: Big data, data science, data mining, machine learning, artificial intelligence, pattern mining, open-source
Open Datasets Yes We also provide over 40 large databases (Kiran, 2022) that can be used with the algorithms offered in PAMI. Furthermore, this library can be beneficial for educational purposes (say, teaching data mining courses) or evaluating algorithms performance. Finally, the website also provides information on how the output of various machine learning libraries can be pipelined into PAMI to discover needy information.
Dataset Splits No The paper describes a software library for pattern mining and provides examples of its usage. It does not present experimental results from specific datasets with defined training/test/validation splits within this paper.
Hardware Specification No The paper mentions support for "heterogeneous computing architectures (e.g., CPU-based, GPU-based, and parallel algorithms based on the mapreduce framework)", but does not provide specific details on hardware models or specifications used for any experimental evaluation within the paper.
Software Dependencies No PAMI is implemented in Python 3.6 and is cross-platform. The source code of PAMI is made available through Git Hub. The execution code of PAMI is made available through the Python Package Index (PYPI) so that users can easily install, update, or delete our library using the pip command.
Experiment Setup No The paper provides an example of how to run an algorithm with a parameter (e.g., `minSup=0.4`), but it does not present a comprehensive experimental setup or specific hyperparameters for experiments conducted within this paper. The paper focuses on describing the library itself rather than presenting new experimental results.