85-Channel多输入和输出通道-卷积神经网络.ipynb

{
 "cells": [
  {
   "attachments": {
    "c4539663-dfc2-40eb-bacc-f1ce9fc046c8.png": {
     "image/png": "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"
    },
    "d6ff1699-f041-4865-9c55-ca4a7195765c.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "id": "c6653299-6f89-4a2b-be05-1b434e3e7895",
   "metadata": {},
   "source": [
    "### 多输入通道\n",
    "![image.png](attachment:d6ff1699-f041-4865-9c55-ca4a7195765c.png)\n",
    "$$(1\\times1+2\\times2+4\\times3+5\\times4)+(0\\times0+1\\times1+3\\times2+4\\times3)=56$$\n",
    "\n",
    "### 多输出通道\n",
    "\n",
    "\n",
    "### 卷积层\n",
    "![image.png](attachment:c4539663-dfc2-40eb-bacc-f1ce9fc046c8.png)\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "2f29fabd-8095-4d90-8c20-2768c5ab8b2b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import dltools\n",
    "\n",
    "# 多输入通道\n",
    "def corr2d_multi_in(X, K):\n",
    "    # 先遍历“X”和“K”的第0个维度（通道维度），再把它们加在一起\n",
    "    return sum(dltools.corr2d(x, k) for x, k in zip(X, K))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "6b293461-dd02-4ed1-8a32-7fb3b271728b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[ 56.,  72.],\n",
       "        [104., 120.]])"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = torch.tensor([[[0.0, 1.0, 2.0], [3.0, 4.0, 5.0], [6.0, 7.0, 8.0]],\n",
    "               [[1.0, 2.0, 3.0], [4.0, 5.0, 6.0], [7.0, 8.0, 9.0]]])\n",
    "K = torch.tensor([[[0.0, 1.0], [2.0, 3.0]], [[1.0, 2.0], [3.0, 4.0]]])\n",
    "\n",
    "corr2d_multi_in(X, K)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "e68b7f4a-fbbb-4d15-bfbd-4ae8b27c37ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 多输出通道\n",
    "def corr2d_multi_in_out(X, K):\n",
    "    # 迭代“K”的第0个维度，每次都对输入“X”执行互相关运算。\n",
    "    # 最后将所有结果都叠加在一起\n",
    "    return torch.stack([corr2d_multi_in(X, k) for k in K], 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "722942ce-f37b-4773-880c-e41a9d6ce8c3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([3, 2, 2, 2])"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "K = torch.stack((K, K + 1, K + 2), 0)\n",
    "K.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "c2321447-846e-4b01-92bc-b5922ea4e110",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[ 56.,  72.],\n",
       "         [104., 120.]],\n",
       "\n",
       "        [[ 76., 100.],\n",
       "         [148., 172.]],\n",
       "\n",
       "        [[ 96., 128.],\n",
       "         [192., 224.]]])"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corr2d_multi_in_out(X, K)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "26b8b628-f7e9-4820-a3ac-f9e7ee1cc778",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 卷积层\n",
    "def corr2d_multi_in_out_1x1(X, K):\n",
    "    c_i, h, w = X.shape\n",
    "    c_o = K.shape[0]\n",
    "    X = X.reshape((c_i, h * w))\n",
    "    K = K.reshape((c_o, c_i))\n",
    "    # 全连接层中的矩阵乘法\n",
    "    Y = torch.matmul(K, X)\n",
    "    return Y.reshape((c_o, h, w))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "807f3bc7-79a5-4150-978b-b7f64ffdfff9",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = torch.normal(0, 1, (3, 3, 3))\n",
    "K = torch.normal(0, 1, (2, 3, 1, 1))\n",
    "\n",
    "Y1 = corr2d_multi_in_out_1x1(X, K)\n",
    "Y2 = corr2d_multi_in_out(X, K)\n",
    "assert float(torch.abs(Y1 - Y2).sum()) < 1e-6"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "87b8b116-b57c-4eef-9870-0802224d8c38",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}