{
    "info": {
        "author": "Daichi Yoshikawa",
        "author_email": "daichi.yoshikawa@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "Development Status :: 4 - Beta",
            "Intended Audience :: Science/Research",
            "License :: OSI Approved :: BSD License",
            "Natural Language :: English",
            "Programming Language :: Python",
            "Programming Language :: Python :: 3.4",
            "Programming Language :: Python :: 3.5",
            "Programming Language :: Python :: 3.6",
            "Programming Language :: Python :: 3.7",
            "Programming Language :: Python :: 3.8",
            "Topic :: Scientific/Engineering :: Artificial Intelligence"
        ],
        "description": "dnnet\n=====\nImplementation of Deep Neural Network with numpy.\nLess dependencies, easier to use.\n\n# Table of Contents\n* Brief tour of dnnet; Introduce small examples, supported methodologies\n* Installation\n* Example; Run sample scripts\n* Use in your project\n\n# Brief tour of dnnet\n## Quick glance of usage\n\nUser can creates instance of NeuralNetwork, add layer one by one,<br>\nfinalize model, set optimizer, execute model fitting, and save model.\n\nIn the below, some arguments are not specified to simplify the example.\n```\nmodel = NeuralNetwork(input_shape=(1, 28, 28), dtype=np.float32)\n\nmodel.add(ConvolutionLayer(filter_shape=(32, 3, 3))\nmodel.add(BatchNormLayer())\nmodel.add(ActivationLayer(activation=Activation.Type.relu))\nmodel.add(DropoutLayer(drop_ratio=0.25))\n\nmodel.add(AffineLayer(output_shape=10)\nmodel.add(ActivationLayer(activation=Activation.Type.softmax)\nmodel.compile()\n\noptimizer = AdaGrad(learning_rate=1e-3, weight_decay=1e-3)\nlearning_curve = model.fit(\n    x=x, y=y, epochs=5, batch=size=100, optimizer=optimizer,\n    loss_function=LossFunction.Type.multinomial_cross_entropy)\nmodel.save(path='./data/output', name='conv_net.dat')\n```\n\nUser can also load model, and predict output.\n```\nmodel.load(path='./data/output', name='conv_net.dat')\ny = model.predict(x)\n```\n\n## Supported Methods\n### Layers\n* Affine\n* Convolution\n* Activation\n* Pool\n* Batch Normalization\n* Dropout\n\n### Activation Functions\n* Sigmoid\n* ReLU\n* ELU\n* Tanh\n* Softmax\n\n### Optimization Methods\n* SGD\n* Momentum\n* AdaGrad\n* Adam\n* AdaDelta\n* RMSProp\n* SMORMS3\n\n### Weight Initialization Methods\n* Xavier's method\n* He's method\n* Default\n\n### Loss Functions\n* MultinomialCrossEntropy for multinomial classification.\n* BinomialCrossEntropy for binary classification.\n* SquaredError for regression.\n\n# Installation\n## Requisites\n* python 3.4 or later\n* numpy 1.12.0 or later\n* matplotlib\n\n## Install dnnet by pip.\n```\npip install dnnet\n```\n\n## Install dnnet from source.\ndnnet doesn't require any complicated path-settings.<br>\nYou just download scripts from github, place it wherever you like,<br>\nand you add some lines like below in your scripts.\n\n```\nimport sys\nsys.path.append('<path-to-dnnet-root-dir>')\n\nfrom dnnet.neuralnet import NeuralNetwork\n```\n\n## Setup environment from scratch (Optional)\nIn this section, setting up python environment from scratch is described.<br>\n\"From scratch\" means that you're supposed to use brand-new computer,<br>\nno python packages (even python itself!) and relevant libraries are installed.\n\nIt may also be useful when you start new python project. In this case,<br>\nyou will partially execute the following steps.\n\n### Setup Python Virtual Environment\n#### Assumption\n* Use python3\n* Make directory for pyenv in \"/home/<user-name>/Documents\"\n* Root directory of your python virtual env is in \"/home/<user-name>/Work/py352_ws\"\n* \"/home/<user-name>/Work/py352_ws/\" is your working directory\n\n#### Setup procedure\n* Install required packages\n```\n$ sudo apt-get install git gcc make openssl libssl-dev libbz2-dev libreadline-dev libsqlite3-dev\n```\n* Install tkinter(This is required to use matplotlib in virtualenv)\n```\n$ sudo apt-get install python3-tk python-tk tk-dev\n```\n* Install pyenv\n```\n   $ cd ~\n   $ git clone git://github.com/yyuu/pyenv.git ./pyenv\n   $ mkdir -p ./pyenv/versions ./pyenv/shims\n```\n* Set paths\nAdd the following description in ~/.bashrc\n```\nexport PYENV_ROOT=${HOME}/Documents/pyenv\nif [ -d \"${PYENV_ROOT}\" ]; then\n  export PATH=${PYENV_ROOT}/bin:$PATH\n  eval \"$(pyenv init -)\"\nfi\n```\nAnd then execute the follows.\n```\n   $ exec $SHELL -l\n   $ . ~/.bashrc\n```\n* Install pyenv-virtualenv\n```\n   $ cd $PYENV_ROOT/plugins\n   $ git clone git://github.com/yyuu/pyenv-virtualenv.git\n```\n* Install python 3.5.2\n\nIf you're going to use theano in the resulting env,\n```\nenv PYTHON_CONFIGURE_OPTS=--enable-shared pyenv install 3.5.2\n```\nelse ...\n```\n   $ pyenv install 3.5.2\n```\n* Setup local pyenv\n```\n   $ mkdir -p ~/Work/py352_ws\n   $ pyenv virtualenv 3.5.2 <name of this environment>\n```\n<name of this environment> can be like py352_env, python3_env, or anything you like.<br>\nHere, it's assumed that you named the environment as \"py352_env\".\n```\n   $ cd ~/Work/py352_ws\n   $ pyenv local py352_env\n   $ pip install --upgrade pip\n```\n\n# Example\n## MNIST\n* Run neural network for mnist.\n```\ncd <path-to-dnnet>/examples/mnist\npython mnist.py\n```\nIf you get an error \"ImportError: Python is not installed as a framework.\",\nit might be because of matplotlib issue.(This happened to me when working with MacOS.)\n\nIn the case, please try the follow.\n```\ncd ~/.matplotlib\necho \"backend: TkAgg\" >> matplotlibrc\n```\n\n# Use in your project\n## If you pip installed dnnet\n```\nfrom dnnet.neuralnet import NeuralNetwork\n```\n\n## If you git cloned dnnet\n```\nimport sys\nsys.path.append('<path-to-dnnet-root-dir>')\n\nfrom dnnet.neuralnet import NeuralNetwork\n```\n\nFor example, if dnnet directory is in ~/Work/dnnet,\ndo like below.\n```\nimport os\nimport sys\nsys.path.append(os.path.join(os.getenv('HOME'), 'Work/dnnet'))\n\nfrom dnnet.neuralnet import NeuralNetwork\n```",
        "description_content_type": "text/markdown",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/daichi-yoshikawa/dnnet",
        "keywords": "",
        "license": "BSD-3-Clause",
        "maintainer": "",
        "maintainer_email": "",
        "name": "dnnet",
        "package_url": "https://pypi.org/project/dnnet/",
        "platform": "",
        "project_url": "https://pypi.org/project/dnnet/",
        "project_urls": {
            "Homepage": "https://github.com/daichi-yoshikawa/dnnet"
        },
        "release_url": "https://pypi.org/project/dnnet/0.9.1/",
        "requires_dist": null,
        "requires_python": ">=3.4",
        "summary": "Deep Neural Network.",
        "version": "0.9.1"
    },
    "last_serial": 4897057,
    "releases": {
        "0.9.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "8cf2968d56563c1a574535bfc81fa492",
                    "sha256": "e4af8dd9a5704e70c7b961ec4f9502d4988d5ab7d6d087279b3aab7413d67c03"
                },
                "downloads": -1,
                "filename": "dnnet-0.9.0.tar.gz",
                "has_sig": false,
                "md5_digest": "8cf2968d56563c1a574535bfc81fa492",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": ">=3.4",
                "size": 22806,
                "upload_time": "2019-03-02T23:19:28",
                "url": "https://files.pythonhosted.org/packages/de/9d/9165add37b7670a69ffc066bd63572dd7e2ff537b8c9987c6c7786b330ac/dnnet-0.9.0.tar.gz"
            }
        ],
        "0.9.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "2dce5184320c34760c0c331b6d460eab",
                    "sha256": "feea5df84ef2773548733bbae655753ce54e8a688c89291a82c717a5d3fa1301"
                },
                "downloads": -1,
                "filename": "dnnet-0.9.1.tar.gz",
                "has_sig": false,
                "md5_digest": "2dce5184320c34760c0c331b6d460eab",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": ">=3.4",
                "size": 23920,
                "upload_time": "2019-03-05T02:11:09",
                "url": "https://files.pythonhosted.org/packages/e0/9f/82f6b3f539e906abc8cb0357212134f4bf2e333f23ffe233187b339ce191/dnnet-0.9.1.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "2dce5184320c34760c0c331b6d460eab",
                "sha256": "feea5df84ef2773548733bbae655753ce54e8a688c89291a82c717a5d3fa1301"
            },
            "downloads": -1,
            "filename": "dnnet-0.9.1.tar.gz",
            "has_sig": false,
            "md5_digest": "2dce5184320c34760c0c331b6d460eab",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": ">=3.4",
            "size": 23920,
            "upload_time": "2019-03-05T02:11:09",
            "url": "https://files.pythonhosted.org/packages/e0/9f/82f6b3f539e906abc8cb0357212134f4bf2e333f23ffe233187b339ce191/dnnet-0.9.1.tar.gz"
        }
    ]
}