{"id":16130951,"url":"https://github.com/rvinas/nnn","last_synced_at":"2025-03-18T14:31:07.031Z","repository":{"id":129460761,"uuid":"86944654","full_name":"rvinas/nnn","owner":"rvinas","description":"Neural network from scratch in Python using Numpy","archived":false,"fork":false,"pushed_at":"2017-05-28T00:20:41.000Z","size":670,"stargazers_count":9,"open_issues_count":0,"forks_count":1,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-02-28T10:08:35.433Z","etag":null,"topics":["deep-learning","deep-learning-library","deep-learning-tutorial","deep-neural-networks","linear-separability","neural-network","numpy","python"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/rvinas.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2017-04-01T21:44:59.000Z","updated_at":"2024-04-28T05:09:17.000Z","dependencies_parsed_at":"2023-04-12T08:16:22.747Z","dependency_job_id":null,"html_url":"https://github.com/rvinas/nnn","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rvinas%2Fnnn","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rvinas%2Fnnn/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rvinas%2Fnnn/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rvinas%2Fnnn/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/rvinas","download_url":"https://codeload.github.com/rvinas/nnn/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":243933420,"owners_count":20370986,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","deep-learning-library","deep-learning-tutorial","deep-neural-networks","linear-separability","neural-network","numpy","python"],"created_at":"2024-10-09T22:18:48.589Z","updated_at":"2025-03-18T14:31:07.026Z","avatar_url":"https://github.com/rvinas.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Numpy Neural Network\nNeural Network from scratch in Python exclusively using Numpy.\n## Overview\n\nThis project consists of a neural network implementation from scratch. Modules are organized in a way that intends to provide both an understandable implementation of neural networks and a user-friendly API.\nThe project is structured as follows:\n- `nnn`\n    - `core/`: Network main functionalities. \n        - `neural_network.py`: Implementation of the neural network, including functionalities such as building a custom network and training it with backpropagation + stochastic gradient descent.\n        - `layers.py`: Defines layers that can take part in the neural network by describing its behavior at the forward and backward steps.\n        - `initializers.py`: Functions used to initialize network's weights.\n        - `activations.py`: Activation functions that may be used to add nonlinearities to the model.\n        - `objectives.py`: Functions to be optimized by adjusting network's parameters.\n    - `utils/`: Other utilities.\n        - `plots.py`: Module with plotting tools. It contains a function to plot the classification boundaries of a 2d classifier, among others.\n        - `loss_grid.py`: Computation of the loss grid for a given neural network and objective function.\n    - `examples`: Simple instructive examples. See [nnn/examples/](./nnn/examples).\n        - `a_greater_than_b.py`: Example demonstrating that a linearly separable dataset can be classified using a neural network without any hidden layer.\n        - `a_aprox_b.py`: Example demonstrating that a non linearly separable dataset requires at least a hidden layer in order to classify samples correctly.\n\n## Prerequisites\n\n- Python 3.5\n- Pip 9.0.1\n\nNote: Not tested with other python versions.\n\n## Installation\nOnce you have met the prerequisites, a single step is required to install this software:\n1. Run `python setup.py install`\n\nThis will install `numpy` (the only required external library to run the neural network) and `matplotlib` (only needed to plot classifier boundaries when running an example).\n\n## Further improvements\n\nThere are several functionalities that may be implemented to make this software more useful:\n- Other types of layers: LSTM, CNN, embeddings,...\n- Batches\n- More optimizers other than Stochastic Gradient Descent\n- More activations\n- More initializers\n- More objective functions\n- Regularization\n- Parallelization\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frvinas%2Fnnn","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Frvinas%2Fnnn","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frvinas%2Fnnn/lists"}