統計推斷(一) Hypothesis Test

個人博客地址 Glooow，歡迎光臨~~~

1. Binary Bayesian hypothesis testing

1.0 Problem Setting

• Hypothesis
  • Hypothesis space $\mathcal{H}=\{H_0, H_1\}$
  • Bayesian approach: Model the valid hypothesis as an RV H
  • Prior $P_0 = p_\mathsf{H}(H_0), P_1=p_\mathsf{H}(H_1)=1-P_0$
• Observation
  • Observation space $\mathcal{Y}$
  • Observation Model $p_\mathsf{y|H}(\cdot|H_0), p_\mathsf{y|H}(\cdot|H_1)$
• Decision rule $f:\mathcal{Y\to H}$
• Cost function $C: \mathcal{H\times H} \to \mathbb{R}$
  • Let $C_{ij}=C(H_j,H_i), correct hypo is H_j$
  • $C$ is valid if $C_{jj}<C_{ij}$
• Optimum decision rule $\hat{H}(\cdot) = \arg\min\limits_{f(\cdot)}\mathbb{E}[C(\mathsf{H},f(\mathsf{y}))]$

1.1 Binary Bayesian hypothesis testing

Theorem: The optimal Bayes’ decision takes the form
$L(\mathsf{y}) \triangleq \frac{p_\mathsf{y|H}(\cdot|H_1)}{p_\mathsf{y|H}(\cdot|H_0)} \overset{H_1} \gtreqless \frac{P_0}{P_1} \frac{C_{10}-C_{00}}{C_{01}-C_{11}} \triangleq \eta$
Proof:
KaTeX parse error: No such environment: align at position 8: \begin{̲a̲l̲i̲g̲n̲}̲ \varphi(f) &=…
Given $y^*$

• if $f(y^*)=H_0$, $\mathbb{E}=C_{00}p_{\mathsf{H|y}}(H_0|y^*)+C_{01}p_{\mathsf{H|y}}(H_1|y^*)$
• if $f(y^*)=H_1$, $\mathbb{E}=C_{10}p_{\mathsf{H|y}}(H_0|y^*)+C_{11}p_{\mathsf{H|y}}(H_1|y^*)$

So
$\frac{p_\mathsf{H|y}(H_1|y^*)}{p_\mathsf{H|y}(H_0|y^*)} \overset{H_1} \gtreqless \frac{C_{10}-C_{00}}{C_{01}-C_{11}}$
備註：證明過程中，注意貝葉斯檢驗爲確定性檢驗，因此對於某個確定的 y，$f(y)=H_1$ 的概率要麼爲 0 要麼爲 1。因此對代價函數求期望時，把 H 看作是隨機變量，而把 $f(y)$ 看作是確定的值來分類討論

Special cases

• Maximum a posteriori (MAP)
  • $C_{00}=C_{11}=0,C_{01}=C_{10}=1$
  • $\hat{H}(y)==\arg\max\limits_{H\in\{H_0,H_1\}} p_\mathsf{H|y}(H|y)$
• Maximum likelihood (ML)
  • $C_{00}=C_{11}=0,C_{01}=C_{10}=1, P_0=P_1=0.5$
  • $\hat{H}(y)==\arg\max\limits_{H\in\{H_0,H_1\}} p_\mathsf{y|H}(y|H)$

1.2 Likelyhood Ratio Test

Generally, LRT
$L(\mathsf{y}) \triangleq \frac{p_\mathsf{y|H}(\cdot|H_1)}{p_\mathsf{y|H}(\cdot|H_0)} \overset{H_1} \gtreqless \eta$

• Bayesian formulation gives a method of calculating $\eta$
• $L(y)$ is a sufficient statistic for the decision problem
• $L(y)$ 的可逆函數也是充分統計量

充分統計量

1.3 ROC

• Detection probability $P_D = P(\hat{H}=H_1 | \mathsf{H}=H_1)$
• False-alarm probability $P_F = P(\hat{H}=H_1 | \mathsf{H}=H_0)$

性質（重要！）

• LRT 的 ROC 曲線是單調不減的

2. Non-Bayesian hypo test

• Non-Bayesian 不需要先驗概率或者代價函數

Neyman-Pearson criterion

$\max_{\hat{H}(\cdot)}P_D \ \ \ s.t. P_F\le \alpha$

Theorem(Neyman-Pearson Lemma)：NP 準則的最優解由 LRT 得到，其中 $\eta$ 由以下公式得到
$P_F=P(L(y)\ge\eta | \mathsf{H}=H_0) = \alpha$
Proof：

物理直觀：同一個 $P_F$ 時 LRT 的 $P_D$ 最大。物理直觀來看，LRT 中判決爲 H1 的區域中 $\frac{p(y|H_1)}{p(y|H_0)}$ 都儘可能大，因此 $P_F$ 相同時 $P_D$ 可最大化

備註：NP 準則最優解爲 LRT，原因是

• 同一個 $P_F$ 時， LRT 的 $P_D$ 最大
• LRT 取不同的 $\eta$ 時，$P_F$ 越大，則 $P_D$ 也越大，即 ROC 曲線單調不減

3. Randomized test

3.1 Decision rule

• Two deterministic decision rules $\hat{H'}(\cdot),\hat{H''}(\cdot)$

• Randomized decision rule $\hat{H}(\cdot)$ by time-sharing
  $\hat{\mathrm{H}}(\cdot)=\left\{\begin{array}{ll}{\hat{H}^{\prime}(\cdot),} & {\text { with probability } p} \\ {\hat{H}^{\prime \prime}(\cdot),} & {\text { with probability } 1-p}\end{array}\right.$

  • Detection prob $P_D=pP_D'+(1-p)P_D''$
  • False-alarm prob $P_F=pP_F'+(1-P)P_F''$

• A randomized decision rule is fully described by $p_{\mathsf{\hat{H}|y}}(H_m|y)$ for m=0,1

3.2 Proposition

1. Bayesian case: cannot achieve a lower Bayes’ risk than the optimum LRT

   Proof: Risk for each y is linear in $p_{\mathrm{H} | \mathbf{y}}\left(H_{0} | \mathbf{y}\right)$, so the minima is achieved at 0 or 1, which degenerate to deterministic decision
   KaTeX parse error: No such environment: align at position 8: \begin{̲a̲l̲i̲g̲n̲}̲ \varphi(\mathb…

2. Neyman-Pearson case:

   1. continuous-valued: For a given $P_F$ constraint, randomized test cannot achieve a larger $P_D$ than optimum LRT
   2. discrete-valued: For a given $P_F$ constraint, randomized test can achieve a larger $P_D$ than optimum LRT. Furthermore, the optimum rand test corresponds to simple time-sharing between the two LRTs nearby

3.3 Efficient frontier

Boundary of region of achievable $(P_D,P_F)$ operation points

• continuous-valued: ROC of LRT
• discrete-valued: LRT points and the straight line segments

Facts

• $P_D \ge P_F$
• efficient frontier is concave function
• $\frac{dP_D}{dP_F}=\eta$

4. Minmax hypo testing

prior: unknown, cost fun: known

4.1 Decision rule

• minmax approach
  $\hat H(\cdot)=\arg\min_{f(\cdot)}\max_{p\in[0,1]} \varphi(f,p)$

• optimal decision rule
  $\hat H(\cdot)=\hat{H}_{p_*}(\cdot) \\ p_* = \arg\max_{p\in[0,1]} \varphi(\hat H_p, p)$

  要想證明上面的最優決策，首先引入 mismatch Bayes decision
  $\hat{\mathrm{H}}_q(y)=\left\{ \begin{array}{ll}{H_1,} & {L(y) \ge \frac{1-q}{q}\frac{C_{10}-C_{00}}{C_{01}-C_{11}}} \\ {H_0,} & {otherwise}\end{array}\right.$
  代價函數如下，可得到 $\varphi(\hat H_q,p)$ 與概率 $p$ 成線性關係
  $\varphi(\hat H_q,p)=(1-p)[C_{00}(1-P_F(q))+C_{10}P_F(q)] + p[C_{01}(1-P_D(q))+C_{11}P_D(q)]$
  Lemma: Max-min inequality
  $\max_x\min_y g(x,y) \le \min_y\max_x g(x,y)$
  Theorem:
  $\min_{f(\cdot)}\max_{p\in[0,1]}\varphi(f,p)=\max_{p\in[0,1]}\min_{f(\cdot)}\varphi(f,p)$
  Proof of Lemma: Let $h(x)=\min_y g(x,y)$
  $\begin{aligned} g(x) &\leq f(x, y), \forall x \forall y \\ \Longrightarrow \max _{x} g(x) & \leq \max _{x} f(x, y), \forall y \\ \Longrightarrow \max _{x} g(x) & \leq \min _{y} \max _{x} f(x, y) \end{aligned}$
  Proof of Thm: 先取 $\forall p_1,p_2 \in [0,1]$，可得到
  $\varphi(\hat H_{p_1},p_1)=\min_f \varphi(f,p_1) \le \max_p \min_f \varphi(f,p) \le \min_f \max_p \varphi(f, p) \le \max_p \varphi(\hat H_{p_2}, p)$
  由於 $p_1,p_2$ 任取時上式都成立，因此可以取 $p_1=p_2=p_*=\arg\max_p \varphi(\hat H_p, p)$

  要想證明定理則只需證明 $\varphi(\hat H_{p_*},p_*)=\max_p \varphi(\hat H_{p_*}, p)$

  由前面可知 $\varphi(\hat H_q,p)$ 與 $p$ 成線性關係，因此要證明上式

  • 若 $p_* \in (0,1)$，只需 $\left.\frac{\partial \varphi\left(\hat{H}_{q^{*}}, p\right)}{\partial p}\right|_{\text {for any } p}=0$，等式自然成立
  • 若 $p_* = 1$，只需 $\left.\frac{\partial \varphi\left(\hat{H}_{q^{*}}, p\right)}{\partial p}\right|_{\text {for any } p} > 0$，最優解就是 $p=1$；$q_*=0$ 同理

  根據下面的引理，可以得到最優決策就是 Bayes 決策 $p_*=\arg\max_p \varphi(\hat H_p, p)$，其中 $p_*$ 滿足
  $\begin{aligned} 0 &=\frac{\partial \varphi\left(\hat{H}_{p_{*}}, p\right)}{\partial p} \\ &=\left(C_{01}-C_{00}\right)-\left(C_{01}-C_{11}\right) P_{\mathrm{D}}\left(p_{*}\right)-\left(C_{10}-C_{00}\right) P_{\mathrm{F}}\left(p_{*}\right) \end{aligned}$
  Lemma:
  $\left.\frac{\mathrm{d} \varphi\left(\hat{H}_{p}, p\right)}{\mathrm{d} p}\right|_{p=q}=\left.\frac{\partial \varphi\left(\hat{H}_{q}, p\right)}{\partial p}\right|_{p=q}=\left.\frac{\partial \varphi\left(\hat{H}_{q}, p\right)}{\partial p}\right|_{\text {for any } p}$
  

其他內容請看：
統計推斷(一) Hypothesis Test
統計推斷(二) Estimation Problem
統計推斷(三) Exponential Family
統計推斷(四) Information Geometry
統計推斷(五) EM algorithm
統計推斷(六) Modeling
統計推斷(七) Typical Sequence
統計推斷(八) Model Selection
統計推斷(九) Graphical models
統計推斷(十) Elimination algorithm
統計推斷(十一) Sum-product algorithm