SQL based Softmax Regression

Intro

Repository for Implementing SQL-based Softmax Regression for MNIST data paper experiment code [paper] [ppt]

Requirements

• Ubuntu 18.04.6 LTS
• Anaconda 23.7.2
• PostgreSQL 16.0

Installation

Python (Hyper API)

1. Install using environment.yml file

$ conda env create -f environment.yml

2. Activate conda environment

$ conda activate sql-4-ml

3. If you want to delete conda environment

$ conda env remove -n sql-4-ml

PostgreSQL

1. Create a user and login to postgres account

$ adduser postgres
$ su - postgres

2. Download the source code from PostgreSQL website

$ wget https://ftp.postgresql.org/pub/source/v16.0/postgresql-16.0.tar.gz
$ tar -xvzf postgresql-16.0.tar.gz

3. Make a directory for build and configure the source tree

$ cd postgresql-16.0
$ mkdir build
$ ./configure --prefix=/home/postgres/postgrsql-16.0/build

4. Build and install the source code (make -j = Make build faster using multiple processes)

$ make -j install

5. Add the shared library path to ~/.bashrc

$ vim ~/.bashrc
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/postgres/postgresql-16.0/build/lib
export PATH=/home/postgres/postgresql-16.0/build/bin:$PATH

$ source ~/.bashrc

• Add the shared library to .profile of postgres user

$ cd ~
$ vim .profile
PATH=$PATH:/home/postgres/postgresql-9.4.5/build/bin
export PATH
$ . ~/.profile

6. Initialize the database storage with initdb command of postgres

$ initdb -D /home/postgres/test_data

How to Execute?

Run PostgreSQL

$ pg_ctl -D /home/postgres/test_data -l logfile start

• Stop the PostgreSQL server

$ pg_ctl -D /home/postgres/test_data -m smart stop

Run Experiment

$ conda activate sql-4-ml
$ python3 experiment.py

References

• https://github.com/mark-blacher/sql-algorithms
• https://github.com/LeeBohyun/postgreSQL/blob/main/postgres-installation-guide.md