264 lines (264 with data), 6.3 kB

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "DeepEEG_Sim.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/kylemath/DeepEEG/blob/master/notebooks/DeepEEG_Sim.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cH7KRd8ZZPMd",
        "colab_type": "text"
      },
      "source": [
        "## DeepEEG\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KjZu4dFMFHJV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!git clone https://github.com/kylemath/DeepEEG\n",
        "!chmod +x ./DeepEEG/install.sh\n",
        "%cd DeepEEG\n",
        "!./install.sh\n",
        "from utils import *\n",
        "%matplotlib inline\n",
        "%cd .."
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FhkDPfIq1ewa",
        "colab_type": "text"
      },
      "source": [
        "#Simulate Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jhU8hlcg1e6T",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "raw,event_id = SimulateRaw(amp1 = 50, amp2 = 5, freq = 2, batch = 2)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AVBtNVebDtUc",
        "colab_type": "text"
      },
      "source": [
        "#Run Preprocessing\n",
        "\n",
        "**Input:  mne.raw, event_id**\n",
        "\n",
        "**Output: mne.epochs**\n",
        "```python\n",
        "plot_psd=False\n",
        "filter_data=True\n",
        "eeg_filter_highpass=1\n",
        "plot_events=False\n",
        "epoch_time=(-.2,1)\n",
        "baseline=(-.2,0)\n",
        "rej_thresh_uV=200\n",
        "rereference=False\n",
        "emcp_raw=False\n",
        "emcp_epochs=False\n",
        "epoch_decim=1\n",
        "plot_electrodes=False\n",
        "plot_erp=False\n",
        "```\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RLIKMk6P453f",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "epochs = PreProcess(raw, event_id,filter_data=False,epoch_time = (-.2,1),plot_erp=True)                \n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iceVBB8vFxcr",
        "colab_type": "text"
      },
      "source": [
        "#Plot Features:\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LqA4ZXaPFxkC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "pick = 33 \n",
        "#select electrode\n",
        "for event in event_id.keys():\n",
        "  fig = plt.imshow(epochs[event]._data[:,pick,:])\n",
        "  plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pwqD_voiDypa",
        "colab_type": "text"
      },
      "source": [
        "#Run FeatureEngineer\n",
        "**Input:  mne.epochs**\n",
        "\n",
        "**Output: deepeeg.feats**\n",
        "```python\n",
        "model_type='NN'\n",
        "frequency_domain=False\n",
        "normalization=True\n",
        "electrode_median=False\n",
        "wavelet_decim=1\n",
        "flims=(3,30)\n",
        "f_bins=20\n",
        "wave_cycles=3\n",
        "spect_baseline=[-1,-.5]\n",
        "electrodes_out=[11,12,13,14,15]\n",
        "test_split = 0.2\n",
        "val_split = 0.2\n",
        "random_seed=1017\n",
        "watermark = False\n",
        "```"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i_jYlTW1A6sb",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "feats = FeatureEngineer(epochs, model_type = 'NN',\n",
        "                        frequency_domain=True, \n",
        "                        normalization= False,\n",
        "                        flims=(5,20),spect_baseline=[-.5,0]\n",
        "                        )"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9ChMvr-jKp8P",
        "colab_type": "text"
      },
      "source": [
        "# Run CreateModel\n",
        "\n",
        "**Input: deepeeg.feats**\n",
        "\n",
        "**Output: deepeeg.model, deepeeg.encoder**\n",
        "\n",
        "```python\n",
        "units=[16,8,4,8,16]\n",
        "dropout=.25\n",
        "batch_norm=True\n",
        "filt_size=3\n",
        "pool_size=2\n",
        "```"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HFf3rBbJKqHR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model, _ = CreateModel(feats, units=[512,512])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-EKBjSapfE4O",
        "colab_type": "text"
      },
      "source": [
        "# TrainTestVal\n",
        "\n",
        "**Input: deepEEG.model, deepEEG.feats**\n",
        "\n",
        "```python\n",
        "batch_size=2\n",
        "train_epochs=20\n",
        "show_plots=True\n",
        "```"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u6ize7eJfB3J",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "TrainTestVal(model, feats)"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}