Day_033_HW-機器如何學習.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 練習時間"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "1. 泛化能力越高代表模型在沒看過的測試資料上的表現能夠越好\n",
    "2. 分類問題可使用交叉商 (Cross Entropy)。回歸問題可使用均方差 (Mean Square Error)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "所謂泛化能力(generalization ability)是指機器學習演算法對新鮮樣本的適應能力。\n",
    "學習的目的是學到隱含在資料對背後的規律，對具有同一規律的學習集以外的資料，經過訓練的網路也能給出合適的輸出，該能力稱為泛化能力。\n",
    "Regularization：減少 Overfitting ，提高模型泛化能力\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "連續損失函數：（A）MSE 損失函數；（B）MAE 損失函數；（C）Huber 損失函數；（D）分位數損失函數。\n",
    "MAE 損失模型的預測結果受脈衝噪聲的影響較小，而 MSE 損失函數的預測結果受此影響略有偏移。\n",
    "均方誤差 (MSE) 是最常用的回歸損失函數，計算方法是求預測值與真實值之間距離的平方和"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "請觀看李宏毅教授以神奇寶貝進化 CP 值預測的範例，解說何謂機器學習與過擬合。並回答以下問題"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[youtube](https://www.youtube.com/watch?v=fegAeph9UaA)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. 模型的泛化能力 (generalization) 是指什麼？ \n",
    "### 2. 分類問題與回歸問題分別可用的目標函數有哪些？"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}