1,592
个编辑
更改
跳到导航
跳到搜索
小第1行：
第1行： − + − + − − − − 第15行：
第11行： − 显示加减关系的文氏图各种信息测量与相关变量数学 x / 数学和 y / 数学相关。两个圆所包含的面积是联合熵 math Eta (x，y) / math。左边的圆圈(红色和紫色)是个体熵数学 Eta (x) / math，红色的是条件熵数学 Eta (x | y) / math。右边的圆(蓝色和紫色)是 math Eta (y) / math，蓝色的是 math Eta (y | x) / math。紫色是互信息 math operatorname { i }(x; y) / math. ]+ + + − In [[information theory]], the '''conditional entropy''' (or '''equivocation''') quantifies the amount of information needed to describe the outcome of a [[random variable]] <math>Y</math> given that the value of another random variable <math>X</math> is known. Here, information is measured in [[Shannon (unit)|shannon]]s, [[Nat (unit)|nat]]s, or [[Hartley (unit)|hartley]]s. The ''entropy of <math>Y</math> conditioned on <math>X</math>'' is written as <math>\Eta(Y|X)</math>.+ − In information theory, the conditional entropy (or equivocation) quantifies the amount of information needed to describe the outcome of a random variable <math>Y</math> given that the value of another random variable <math>X</math> is known. Here, information is measured in shannons, nats, or hartleys. The entropy of <math>Y</math> conditioned on <math>X</math> is written as <math>\Eta(Y|X)</math>. − 在信息论中，假设另一个随机变量 math x / math 的值是已知的，信息条件熵量化描述随机变量 math y / math 的结果所需的信息量。在这里，信息是用夏农、纳特斯或哈特利来衡量的。数学 y / 数学的熵以数学 x / 数学为条件，表示为数学 Eta (y | x) / 数学。 − − − − − − == Definition == − − 定义 第41行：
第29行： − 数学 y / 数学给定数学 x / 数学的条件熵定义为+ − − 第57行：
第43行： − 不会有事的+ 第63行：
第49行： − 标题+ 第69行：
第55行： − 会公式开始+ 第75行：
第61行： − 6号手术室+ 第93行：
第79行： − + − − 第103行：
第87行： − 其中 math mathcal x / math 和 math mathcal y / math 表示数学 x / math 和数学 y / math 的支持集。+ − − 第111行：
第93行： − + − − 注意: 对于固定数学 c 0 / math，表达式 math 0 log 0 / math 和 math 0 log c / 0 / math 应当被视为等于零。这是因为 math theta 0 ^ + theta log theta 0 / math 和 math theta 0 ^ + theta log theta 0 / math！ -- 因为 p (x，y)仍然可以等于0，即使 p (x) ！ 0和 p (y) ！ 0.P (x，y) p (x)0怎么样？--> − + 第121行：
第101行： − Intuitive explanation of the definition : + − − 对定义的直观解释: − − According to the definition, <math>\displaystyle H( Y|X) =\mathbb{E}( \ f( X,Y) \ )</math> where <math>\displaystyle f:( x,y) \ \rightarrow -\log_{2}( \ p( y|x) \ ) .</math> <math>\displaystyle f</math> associates to <math>\displaystyle ( x,y)</math> the information content of <math>\displaystyle ( Y=y)</math> given <math>\displaystyle (X=x)</math>, which is the amount of information needed to describe the event <math>\displaystyle (Y=y)</math> given <math>(X=x)</math>. According to the law of large numbers, <math>\displaystyle H(Y|X)</math> is the arithmetic mean of a large number of independent realizations of <math>\displaystyle f(X,Y)</math>. − − According to the definition, <math>\displaystyle H( Y|X) =\mathbb{E}( \ f( X,Y) \ )</math> where <math>\displaystyle f:( x,y) \ \rightarrow -\log_{2}( \ p( y|x) \ ) .</math> <math>\displaystyle f</math> associates to <math>\displaystyle ( x,y)</math> the information content of <math>\displaystyle ( Y=y)</math> given <math>\displaystyle (X=x)</math>, which is the amount of information needed to describe the event <math>\displaystyle (Y=y)</math> given <math>(X=x)</math>. According to the law of large numbers, <math>\displaystyle H(Y|X)</math> is the arithmetic mean of a large number of independent realizations of <math>\displaystyle f(X,Y)</math>. − − 根据定义，math displaystyle h (y | x) mathbb { e }( f (x，y)) / math displaystyle f: (x，y) righttarrow log {2}( p (y | x))。 / math displaystyle f / math 联想到 math displaystyle (x，y) / math 数学数学 displaystyle (y) / math 给定的 math displaystyle (x) / math，这是描述事件数学 displaystyle (y) / math 给定的 math (x) / math 所需的信息量。根据大数定律，math displaystyle h (y | x) / math 是 math displaystyle f (x，y) / math 的大量独立实现的算术平均数。 − − + + + + − + − 动机+ − Let <math>\Eta(Y|X=x)</math> be the entropy of the discrete random variable <math>Y</math> conditioned on the discrete random variable <math>X</math> taking a certain value <math>x</math>. Denote the support sets of <math>X</math> and <math>Y</math> by <math>\mathcal X</math> and <math>\mathcal Y</math>. Let <math>Y</math> have probability mass function <math>p_Y{(y)}</math>. The unconditional entropy of <math>Y</math> is calculated as <math>\Eta(Y) := \mathbb{E}[\operatorname{I}(Y)]</math>, i.e.+ − 设数学是离散型随机变量数学 y / math 的熵，条件是离散型随机变量数学 x / math 取一定值数学 x / math。用 math mathcal x / math 和 math mathcal y / math 表示数学 x / math 和数学 y / math 的支持集。让数学 y / 数学有概率质量函数 / 数学 p {(y)} / 数学。数学 y / math 的无条件熵计算为 math Eta (y) : mathbb { e }[ operatorname { i }(y)] / math，即。+ + + − <math>\Eta(Y) = \sum_{y\in\mathcal Y} {\mathrm{Pr}(Y=y)\,\mathrm{I}(y)} + − 数学中的 Eta (y)(y)(y)(y)(y)(y)(y)+ − + − - 和数学 y }{ py (y) log 2{ py (y)} ，/ math+ + + − − where <math>\operatorname{I}(y_i)</math> is the information content of the outcome of <math>Y</math> taking the value <math>y_i</math>. The entropy of <math>Y</math> conditioned on <math>X</math> taking the value <math>x</math> is defined analogously by conditional expectation: − − 其中 math operatorname { i }(yi) / math 是取值 math y / math 的数学 y / math 结果的信息内容。数学 y / 数学的熵取决于数学 x / 数学的取值，数学 x / 数学的定义类似于条件期望的定义: + + − <math>\Eta(Y|X=x)+ − + − = -\sum_{y\in\mathcal Y} {\Pr(Y = y|X=x) \log_2{\Pr(Y = y|X=x)}}. </math>+ − = -\sum_{y\in\mathcal Y} {\Pr(Y = y|X=x) \log_2{\Pr(Y = y|X=x)}}. </math>+ − − - 和数学 y }{ Pr (y | x) log 2{ Pr (y | x)}。数学 − − − − − − − − − − 数学是对所有可能的数值求平均值的结果，数学 x / 数学可能需要。 + + 第209行：
第175行： − Given discrete random variables <math>X</math> with image <math>\mathcal X</math> and <math>Y</math> with image <math>\mathcal Y</math>, the conditional entropy of <math>Y</math> given <math>X</math> is defined as the weighted sum of <math>\Eta(Y|X=x)</math> for each possible value of <math>x</math>, using <math>p(x)</math> as the weights:+ − − 给定离散随机变量数学 x / 数学 x / 数学 x / 数学 y / 数学 y / 数学，数学 y / 数学 x / 数学的条件熵定义为数学 x / 数学每个可能值的加权和，用数学 p (x) / 数学作为权重: − + − − <math> − − 数学 − \begin{align}+ − Begin { align }+ − + − − 数学 x 中的 Eta (y | x) ，p (x) ，Eta (y | x) − & =-\sum_{x\in\mathcal X} p(x)\sum_{y\in\mathcal Y}\,p(y|x)\,\log\, p(y|x)\\+ − − 数学 x } p (x) sum y } ，p (y | x) ，log，p (y | x) − & =-\sum_{x\in\mathcal X}\sum_{y\in\mathcal Y}\,p(x,y)\,\log\,p(y|x)\\+ − − 数学中的 x 和 y，p (x，y) ，log，p (y | x) − & =-\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log\,p(y|x)\\+ − − 数学 x，y = p (x，y) log，p (y | x) − & =-\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \frac {p(x,y)} {p(x)}. \\+ − (x，y) log-frac { p (x，y)}{ p (x，y)}.\\+ − − & = \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \frac {p(x)} {p(x,y)}. \\ − − (x，y) log frac { p (x)}{ p (x，y)}.\\ − + − End { align }+ − − </math> − − 数学 − − 第283行：
第225行： − <!-- This paragraph is incorrect; the last line is not the KL divergence between any two distributions, since p(x) is [in general] not a valid distribution over the domains of X and Y. The last formula above is the Kullback-Leibler divergence, also known as relative entropy. Relative entropy is always positive, and vanishes if and only if <math>p(x,y) = p(x)</math>. This is when knowing <math>x</math> tells us everything about <math>y</math>. ADDED: Could this comment be out of date since the KL divergence is not mentioned above? November 2014 -->+ − + + + − ==Properties==+ − 属性+ − + − 条件熵等于零+ − <math>\Eta(Y|X)=0</math> if and only if the value of <math>Y</math> is completely determined by the value of <math>X</math>.+ − Math Eta (y | x)0 / math 当且仅当 math y / math 的值完全由 math x / math 的值决定。+ + + − ===Conditional entropy of independent random variables===+ − 独立随机变量的条件熵+ − − Conversely, <math>\Eta(Y|X) = \Eta(Y)</math> if and only if <math>Y</math> and <math>X</math> are independent random variables. − − 相反，math Eta (y | x) Eta (y) / math 当且仅当 math y / math 和 math x / math 是独立随机变量。 + + − − ===Chain rule=== − − 链式规则 − Assume that the combined system determined by two random variables <math>X</math> and <math>Y</math> has joint entropy <math>\Eta(X,Y)</math>, that is, we need <math>\Eta(X,Y)</math> bits of information on average to describe its exact state. Now if we first learn the value of <math>X</math>, we have gained <math>\Eta(X)</math> bits of information. Once <math>X</math> is known, we only need <math>\Eta(X,Y)-\Eta(X)</math> bits to describe the state of the whole system. This quantity is exactly <math>\Eta(Y|X)</math>, which gives the chain rule of conditional entropy:+ − − 假设由两个随机变量数学 x / math 和数学 y / math 组成的组合系统具有联合熵数学 Eta (x，y) / math，也就是说，我们平均需要 math Eta (x，y) / math 位信息来描述它的精确状态。现在，如果我们首先学习数学 x / math 的值，我们就得到了数学 Eta (x) / 数学信息位。一旦知道了数学 x / math，我们只需要 math Eta (x，y)- Eta (x) / math 位来描述整个系统的状态。这个量正是 math Eta (y | x) / math，它给出了条件熵的链式法则: − + 第347行：
第285行： − + − − Math Eta (y | x) ， Eta (x，y)- Eta (x) . / math − + 第357行：
第293行： − The chain rule follows from the above definition of conditional entropy:+ − − 链式规则遵循以上条件熵的定义: − + − − <math>\begin{align} − − 数学 begin { align } − \Eta(Y|X) &= \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \left(\frac{p(x)}{p(x,y)} \right) \\[4pt]+ − Eta (y | x) & sum (x，y) p (x，y) log 左(frac (x)} p (x，y)右)[4 pt ]+ − &= -\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log (p(x,y)) + \sum_{x\in\mathcal X, y\in\mathcal Y}{p(x,y)\log(p(x))} \\[4pt]+ − − 数学 x，y 中数学 y } p (x，y) log (p (x，y)) + 数学 x，y 中数学 y } p (x，y) log (p (x))[4 pt ] − & = \Eta(X,Y) + \sum_{x \in \mathcal X} p(x)\log (p(x) ) \\[4pt]+ − & Eta (x，y) + sum { x } p (x) log (p (x))[4 pt ]+ − & = \Eta(X,Y) - \Eta(X). + − & Eta (x，y)- Eta (x).+ − \end{align}</math>+ − End { align } / math+ + + − In general, a chain rule for multiple random variables holds:+ − 一般来说，多个随机变量的链式规则适用于:+ + + − <math> \Eta(X_1,X_2,\ldots,X_n) =+ − Math Eta (x1，x2， ldots，xn)+ − \sum_{i=1}^n \Eta(X_i | X_1, \ldots, X_{i-1}) </math>+ − − { i } ^ n Eta (xi | x1，ldots，x { i-1}) / math − + 第433行：
第363行： − It has a similar form to chain rule in probability theory, except that addition instead of multiplication is used. − 它有一个类似的形式链规则在概率论，除了加法代替乘法是使用。 + − + − − − ===Bayes' rule=== − − 贝叶斯规则 − Bayes' rule for conditional entropy states+ − 条件熵的贝叶斯规则+ − + − Math Eta (y | x) ， Eta (x | y)- Eta (x) + Eta (y) . / math+ + + − − Proof. <math>\Eta(Y|X) = \Eta(X,Y) - \Eta(X)</math> and <math>\Eta(X|Y) = \Eta(Y,X) - \Eta(Y)</math>. Symmetry entails <math>\Eta(X,Y) = \Eta(Y,X)</math>. Subtracting the two equations implies Bayes' rule. − − 证据。Math Eta (y | x) Eta (x，y)- Eta (x) / math Eta (x | y) Eta (y，x)- Eta (y) / math.对称性需要数学 Eta (x，y) Eta (y，x) / 数学。减去这两个方程就得到了贝叶斯定律。 + + − If <math>Y</math> is conditionally independent of <math>Z</math> given <math>X</math> we have:+ − − 如果数学 y / 数学是条件独立于数学 z / 数学给定的数学 x / 数学，我们有: − + 第485行：
第407行： − + − Math Eta (y | x，z) ， Eta (y | x) . / math+ − − − − ===Other properties=== − − 其他物业 − + − 对于任何数学 x / 数学 y / 数学:+ − + − 数学显示“ block” begin { align }+ − \Eta(Y|X) &\le \Eta(Y) \, \\+ − 三、 Eta (y | x)和 le Eta (y) ,+ − − \Eta(X,Y) &= \Eta(X|Y) + \Eta(Y|X) + \operatorname{I}(X;Y),\qquad \\ − − (x，y) & Eta (x | y) + Eta (y | x) + operatorname { i }(x; y) , − \Eta(X,Y) &= \Eta(X) + \Eta(Y) - \operatorname{I}(X;Y),\, \\+ − Eta (x，y) & Eta (x) + Eta (y)-操作者名称{ i }(x; y) ， ，+ − − \operatorname{I}(X;Y) &\le \Eta(X),\, − − { i }(x; y) & le Eta (x) ， , − − \end{align}</math> − − End { align } / math + + − − where <math>\operatorname{I}(X;Y)</math> is the mutual information between <math>X</math> and <math>Y</math>. − − 其中 math operatorname { i }(x; y) / math 是 math x / math 和 math y / math 之间的相互信息。 − − − − For independent <math>X</math> and <math>Y</math>: − − 对于独立数学 x / 数学 y / 数学: − − − − <math>\Eta(Y|X) = \Eta(Y) </math> and <math>\Eta(X|Y) = \Eta(X) \, </math> − − Math Eta (y | x) Eta (y) / math Eta (x | y) Eta (x) ，/ math − − 第577行：
第465行： − Although the specific-conditional entropy <math>\Eta(X|Y=y)</math> can be either less or greater than <math>\Eta(X)</math> for a given random variate <math>y</math> of <math>Y</math>, <math>\Eta(X|Y)</math> can never exceed <math>\Eta(X)</math>. − 虽然对于给定的随机变量 y / 数学 y / 数学，特定条件熵数学 Eta (x | y) / 数学可以比 math Eta (x) / 数学更小或更大，math Eta (x | y) / 数学永远不能超过 math Eta (x) / 数学。 − − − − − − == Conditional differential entropy == − − 条件微分熵 − − === Definition === − − 定义 − − The above definition is for discrete random variables. The continuous version of discrete conditional entropy is called conditional differential (or continuous) entropy. Let <math>X</math> and <math>Y</math> be a continuous random variables with a joint probability density function <math>f(x,y)</math>. The differential conditional entropy <math>h(X|Y)</math> is defined as − − 上述定义适用于离散型随机变量。离散条件熵的连续形式称为条件微分(或连续)熵。让数学 x / math 和数学 y / math 是一个连续的随机变量和一个概率密度函数 / 数学 f (x，y) / math。微分 / 条件熵数学 h (x | y) / math 定义为 − − − − {{Equation box 1 − − {方程式方框1 − − |indent = − − |indent = − − 不会有事的 − − |title= − − |title= − − 标题 − − |equation = {{NumBlk||<math>h(X|Y) = -\int_{\mathcal X, \mathcal Y} f(x,y)\log f(x|y)\,dx dy</math>|{{EquationRef|Eq.2}}}} − − |equation = }} − − 会公式开始 − − |cellpadding= 6 − − |cellpadding= 6 − − 6号手术室 − − |border − − |border − − 边界 − − |border colour = #0073CF − − |border colour = #0073CF − − 0073CF − − |background colour=#F5FFFA}} − − |background colour=#F5FFFA}} − − 5 / fffa } − − − − − − === Properties === − − === Properties === − − 属性 − − In contrast to the conditional entropy for discrete random variables, the conditional differential entropy may be negative. − − In contrast to the conditional entropy for discrete random variables, the conditional differential entropy may be negative. − − 与离散随机变量的条件熵相反，条件微分熵可能是负的。 − − − − − − As in the discrete case there is a chain rule for differential entropy: − − As in the discrete case there is a chain rule for differential entropy: − − 在离散情况下，微分熵有一个链式规则: − − :<math>h(Y|X)\,=\,h(X,Y)-h(X)</math><ref name=cover1991 />{{rp|253}} − − <math>h(Y|X)\,=\,h(X,Y)-h(X)</math> − − 数学 h (y | x) ，h (x，y)-h (x) / math − − Notice however that this rule may not be true if the involved differential entropies do not exist or are infinite. − − Notice however that this rule may not be true if the involved differential entropies do not exist or are infinite. − − 然而，请注意，如果所涉及的微分熵不存在或者是无限的，那么这个规则可能不成立。 − − − − − − Joint differential entropy is also used in the definition of the [[mutual information]] between continuous random variables: − − Joint differential entropy is also used in the definition of the mutual information between continuous random variables: − − 联合微分熵也用于连续随机变量之间互信息的定义: − − :<math>\operatorname{I}(X,Y)=h(X)-h(X|Y)=h(Y)-h(Y|X)</math> − − <math>\operatorname{I}(X,Y)=h(X)-h(X|Y)=h(Y)-h(Y|X)</math> − − { i }(x，y) h (x)-h (x | y) h (y)-h (y | x) / math − − − − − − <math>h(X|Y) \le h(X)</math> with equality if and only if <math>X</math> and <math>Y</math> are independent.<ref name=cover1991 />{{rp|253}} − − <math>h(X|Y) \le h(X)</math> with equality if and only if <math>X</math> and <math>Y</math> are independent. − − 数学 h (x | y) le h (x) / math with equality 当且仅当数学 x / math 和数学 y / math 是独立的。 − − − − − − ===Relation to estimator error=== − − ===Relation to estimator error=== − − 与估计误差的关系 − − The conditional differential entropy yields a lower bound on the expected squared error of an [[estimator]]. For any random variable <math>X</math>, observation <math>Y</math> and estimator <math>\widehat{X}</math> the following holds:<ref name=cover1991 />{{rp|255}} − − The conditional differential entropy yields a lower bound on the expected squared error of an estimator. For any random variable <math>X</math>, observation <math>Y</math> and estimator <math>\widehat{X}</math> the following holds: − − 条件微分熵对估计量的期望平方误差产生一个下限。对于任何随机变量的数学 x / math，观察数学 y / math 和估计数学 x / math，下面的观点成立: − − :<math display="block">\mathbb{E}\left[\bigl(X - \widehat{X}{(Y)}\bigr)^2\right] − − <math display="block">\mathbb{E}\left[\bigl(X - \widehat{X}{(Y)}\bigr)^2\right] − − 数学显示块“左”[ bigl (x-widehat {(y)} bigr) ^ 2] − − \ge \frac{1}{2\pi e}e^{2h(X|Y)}</math> − − \ge \frac{1}{2\pi e}e^{2h(X|Y)}</math> − − (x | y)} / math − − − − − − This is related to the [[uncertainty principle]] from [[quantum mechanics]]. − − This is related to the uncertainty principle from quantum mechanics. − − 这与量子力学的不确定性原理有关。 − − − − − − ==Generalization to quantum theory== − − ==Generalization to quantum theory== − − 对量子理论的推广 − − In [[quantum information theory]], the conditional entropy is generalized to the [[conditional quantum entropy]]. The latter can take negative values, unlike its classical counterpart. − − In quantum information theory, the conditional entropy is generalized to the conditional quantum entropy. The latter can take negative values, unlike its classical counterpart. − − 在量子信息论中，条件熵被推广为条件量子熵。后者可以采取负值，不像它的古典对应物。 − − − − − − == See also == − − == See also == − − 参见 − − * [[Entropy (information theory)]] − − − − * [[Mutual information]] − − − − * [[Conditional quantum entropy]] − − − − * [[Variation of information]] − − − − * [[Entropy power inequality]] − − − − * [[Likelihood function]] − − − − − − − − ==References== − − ==References== − − 参考资料 − − {{Reflist}} − − − − − − − − [[Category:Entropy and information]] 第831行：
第481行： − [[Category:Information theory]]+
Moved page from wikipedia:en:Conditional entropy (history)
此词条暂由彩云小译翻译，未经人工整理和审校，带来阅读不便，请见谅。{{Short description|Measure of relative information in probability theory}}
此词条暂由彩云小译翻译，未经人工整理和审校，带来阅读不便，请见谅。
{{Short description|Measure of relative information in probability theory}}
{{Information theory}}
{{Information theory}}
[[Venn diagram showing additive and subtractive relationships various information measures associated with correlated variables <math>X</math> and <math>Y</math>. The area contained by both circles is the joint entropy <math>\Eta(X,Y)</math>. The circle on the left (red and violet) is the individual entropy <math>\Eta(X)</math>, with the red being the conditional entropy <math>\Eta(X|Y)</math>. The circle on the right (blue and violet) is <math>\Eta(Y)</math>, with the blue being <math>\Eta(Y|X)</math>. The violet is the mutual information <math>\operatorname{I}(X;Y)</math>.]]
[[Venn diagram showing additive and subtractive relationships various information measures associated with correlated variables <math>X</math> and <math>Y</math>. The area contained by both circles is the joint entropy <math>\Eta(X,Y)</math>. The circle on the left (red and violet) is the individual entropy <math>\Eta(X)</math>, with the red being the conditional entropy <math>\Eta(X|Y)</math>. The circle on the right (blue and violet) is <math>\Eta(Y)</math>, with the blue being <math>\Eta(Y|X)</math>. The violet is the mutual information <math>\operatorname{I}(X;Y)</math>.]]
文恩图显示了相加和相减的关系，各种信息测量与相关变量相关。两个圆圈所包含的区域是联合熵。左边的圆圈(红色和紫色)代表个体熵。左边的圆圈代表个体熵。右边的圆圈代表个体熵。右边的圆圈代表个体熵。右边的圆圈代表个体熵。右边的圆圈代表个体熵。右边的圆圈代表个体熵。右边的圆圈代表个体熵。右边的圆圈代表个体熵。右边的圆圈代表个体熵。右边的圆圈代表个体。右边的圆圈(蓝色和紫色)是 < math > Eta (y) </math > ，蓝色的是 < math > Eta (y | x) </math > 。紫色是共同的信息[ math > 操作者名称{ i }(x; y) </math > ]
In [[information theory]], the '''conditional entropy''' quantifies the amount of information needed to describe the outcome of a [[random variable]] <math>Y</math> given that the value of another random variable <math>X</math> is known. Here, information is measured in [[Shannon (unit)|shannon]]s, [[Nat (unit)|nat]]s, or [[Hartley (unit)|hartley]]s. The ''entropy of <math>Y</math> conditioned on <math>X</math>'' is written as <math>\Eta(Y|X)</math>.
In information theory, the conditional entropy quantifies the amount of information needed to describe the outcome of a random variable <math>Y</math> given that the value of another random variable <math>X</math> is known. Here, information is measured in shannons, nats, or hartleys. The entropy of <math>Y</math> conditioned on <math>X</math> is written as <math>\Eta(Y|X)</math>.
在信息论中，如果另一个随机变量的值是已知的，那么条件熵就会量化描述一个随机变量的结果所需的信息量。在这里，信息是用夏农、纳特斯或哈特利来衡量的。“数学”的熵取决于“数学” ，“ x”表示“数学” ，“埃塔”表示“数学”。
== Definition ==
== Definition ==
The conditional entropy of <math>Y</math> given <math>X</math> is defined as
The conditional entropy of <math>Y</math> given <math>X</math> is defined as
The conditional entropy of <math>Y</math> given <math>X</math> is defined as
The conditional entropy of <math>Y</math> given <math>X</math> is defined as
给定的 x 条件熵被定义为
|indent =
|indent =
2012年10月22日
|title=
|title=
|title=
|title=
2012年10月11日
|equation = {{NumBlk||<math>\Eta(Y|X)\ = -\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \frac {p(x,y)} {p(x)}</math>|{{EquationRef|Eq.1}}}}
|equation = {{NumBlk||<math>\Eta(Y|X)\ = -\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \frac {p(x,y)} {p(x)}</math>|{{EquationRef|Eq.1}}}}
|equation = }}
|equation = }}
| equation = }
|cellpadding= 6
|cellpadding= 6
|cellpadding= 6
|cellpadding= 6
6
|border
|border
|background colour=#F5FFFA}}
|background colour=#F5FFFA}}
5 / fffa }
5/fffa }}
where <math>\mathcal X</math> and <math>\mathcal Y</math> denote the support sets of <math>X</math> and <math>Y</math>.
where <math>\mathcal X</math> and <math>\mathcal Y</math> denote the support sets of <math>X</math> and <math>Y</math>.
这里 < math > 数学 x </math > 和 < math > > 数学 y </math > 表示 < math > x </math > 和 < math > y </math > 的支持集。
''Note:'' It is conventioned that the expressions <math>0 \log 0</math> and <math>0 \log c/0</math> for fixed <math>c > 0</math> should be treated as being equal to zero. This is because <math>\lim_{\theta\to0^+} \theta\, \log \,c/\theta = 0</math> and <math>\lim_{\theta\to0^+} \theta\, \log \theta = 0</math><ref>{{Cite web|url=http://www.inference.org.uk/mackay/itprnn/book.html|title=David MacKay: Information Theory, Pattern Recognition and Neural Networks: The Book|website=www.inference.org.uk|access-date=2019-10-25}}</ref> <!-- because p(x,y) could still equal 0 even if p(x) != 0 and p(y) != 0. What about p(x,y)=p(x)=0? -->
''Note:'' It is conventioned that the expressions <math>0 \log 0</math> and <math>0 \log c/0</math> for fixed <math>c > 0</math> should be treated as being equal to zero. This is because <math>\lim_{\theta\to0^+} \theta\, \log \,c/\theta = 0</math> and <math>\lim_{\theta\to0^+} \theta\, \log \theta = 0</math><ref>{{Cite web|url=http://www.inference.org.uk/mackay/itprnn/book.html|title=David MacKay: Information Theory, Pattern Recognition and Neural Networks: The Book|website=www.inference.org.uk|access-date=2019-10-25}}</ref> <!-- because p(x,y) could still equal 0 even if p(x) != 0 and p(y) != 0. What about p(x,y)=p(x)=0? -->
Note: It is conventioned that the expressions <math>0 \log 0</math> and <math>0 \log c/0</math> for fixed <math>c > 0</math> should be treated as being equal to zero. This is because <math>\lim_{\theta\to0^+} \theta\, \log \,c/\theta = 0</math> and <math>\lim_{\theta\to0^+} \theta\, \log \theta = 0</math> <!-- because p(x,y) could still equal 0 even if p(x) != 0 and p(y) != 0. What about p(x,y)=p(x)=0? -->
Note: It is conventioned that the expressions <math>0 \log 0</math> and <math>0 \log c/0</math> for fixed <math>c > 0</math> should be treated as being equal to zero. This is because <math>\lim_{\theta\to0^+} \theta\, \log \,c/\theta = 0</math> and <math>\lim_{\theta\to0^+} \theta\, \log \theta = 0</math>
注意: 常规的表达式 < math > 0 log 0 </math > 和 < math > 0 log c/0 </math > 对于 fixed < math > c > 0 </math > 应该被视为等于零。这是因为 < math > lim { theta to0 ^ + } theta，log，c/theta = 0 </math > 和 < math > lim { theta to0 ^ + } theta，log theta = 0 </math >
Intuitive explanation of the definition :
Intuitive explanation of the definition :
The chain rule follows from the above definition of conditional entropy:
链式规则遵循了上述条件熵的定义:
According to the definition, <math>\displaystyle H( Y|X) =\mathbb{E}( \ f( X,Y) \ )</math> where <math>\displaystyle f:( x,y) \ \rightarrow -\log( \ p( y|x) \ ) .</math> <math>\displaystyle f</math> associates to <math>\displaystyle ( x,y)</math> the information content of <math>\displaystyle ( Y=y)</math> given <math>\displaystyle (X=x)</math>, which is the amount of information needed to describe the event <math>\displaystyle (Y=y)</math> given <math>(X=x)</math>. According to the law of large numbers, <math>\displaystyle H(Y|X)</math> is the arithmetic mean of a large number of independent realizations of <math>\displaystyle f(X,Y)</math>.
<math>\begin{align}
1.1.1.2.2.2.2.2.2.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.4.3.3.3.3.3.3.3.3.3.3.3.4.3.3.3.3.3.3.3.3.3
== Motivation ==
== Motivation ==
== Motivation ==
\Eta(Y|X) &= \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \left(\frac{p(x)}{p(x,y)} \right) \\[4pt]
Eta (y | x) & = sum _ { x in mathcal x，y in mathcal y } p (x，y) log left (frac { p (x)}{ p (x，y)} right)[4 pt ]
Let <math>\Eta(Y|X=x)</math> be the [[Shannon Entropy|entropy]] of the discrete random variable <math>Y</math> conditioned on the discrete random variable <math>X</math> taking a certain value <math>x</math>. Denote the support sets of <math>X</math> and <math>Y</math> by <math>\mathcal X</math> and <math>\mathcal Y</math>. Let <math>Y</math> have [[probability mass function]] <math>p_Y{(y)}</math>. The unconditional entropy of <math>Y</math> is calculated as <math>\Eta(Y) := \mathbb{E}[\operatorname{I}(Y)]</math>, i.e.
Let <math>\Eta(Y|X=x)</math> be the [[Shannon Entropy|entropy]] of the discrete random variable <math>Y</math> conditioned on the discrete random variable <math>X</math> taking a certain value <math>x</math>. Denote the support sets of <math>X</math> and <math>Y</math> by <math>\mathcal X</math> and <math>\mathcal Y</math>. Let <math>Y</math> have [[probability mass function]] <math>p_Y{(y)}</math>. The unconditional entropy of <math>Y</math> is calculated as <math>\Eta(Y) := \mathbb{E}[\operatorname{I}(Y)]</math>, i.e.
&= \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)(\log (p(x))-\log (p(x,y))) \\[4pt]
& = sum _ { x in mathcal x，y in mathcal y } p (x，y)(log (p (x))-log (p (x，y)))[4 pt ]
&= -\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log (p(x,y)) + \sum_{x\in\mathcal X, y\in\mathcal Y}{p(x,y)\log(p(x))} \\[4pt]
& =-sum _ { x in mathcal x，y in mathcal y } p (x，y) log (p (x，y)) + sum _ { x in mathcal x，y in mathcal y }{ p (x，y) log (p (x))}[4 pt ]
:<math>\Eta(Y) = \sum_{y\in\mathcal Y} {\mathrm{Pr}(Y=y)\,\mathrm{I}(y)}
:<math>\Eta(Y) = \sum_{y\in\mathcal Y} {\mathrm{Pr}(Y=y)\,\mathrm{I}(y)}
& = \Eta(X,Y) + \sum_{x \in \mathcal X} p(x)\log (p(x) ) \\[4pt]
& = Eta (x，y) + sum _ { x in mathcal x } p (x) log (p (x))[4 pt ]
= -\sum_{y\in\mathcal Y} {p_Y(y) \log_2{p_Y(y)}},</math>
= -\sum_{y\in\mathcal Y} {p_Y(y) \log_2{p_Y(y)}},</math>
= -\sum_{y\in\mathcal Y} {p_Y(y) \log_2{p_Y(y)}},</math>
& = \Eta(X,Y) - \Eta(X).
& = Eta (x，y)-Eta (x).
\end{align}</math>
结束{ align } </math >
where <math>\operatorname{I}(y_i)</math> is the [[information content]] of the [[Outcome (probability)|outcome]] of <math>Y</math> taking the value <math>y_i</math>. The entropy of <math>Y</math> conditioned on <math>X</math> taking the value <math>x</math> is defined analogously by [[conditional expectation]]:
where <math>\operatorname{I}(y_i)</math> is the [[information content]] of the [[Outcome (probability)|outcome]] of <math>Y</math> taking the value <math>y_i</math>. The entropy of <math>Y</math> conditioned on <math>X</math> taking the value <math>x</math> is defined analogously by [[conditional expectation]]:
In general, a chain rule for multiple random variables holds:
一般来说，多个随机变量的链式规则适用于:
:<math>\Eta(Y|X=x)
:<math>\Eta(Y|X=x)
= -\sum_{y\in\mathcal Y} {\Pr(Y = y|X=x) \log_2{\Pr(Y = y|X=x)}}.</math>
(y | x)
<math> \Eta(X_1,X_2,\ldots,X_n) =
< math > Eta (x1，x2，ldots，xn) =
Note that <math>\Eta(Y|X)</math> is the result of averaging <math>\Eta(Y|X=x)</math> over all possible values <math>x</math> that <math>X</math> may take. Also, if the above sum is taken over a sample <math>y_1, \dots, y_n</math>, the expected value <math>E_X[ \Eta(y_1, \dots, y_n \mid X = x)]</math> is known in some domains as '''equivocation'''.<ref>{{cite journal|author1=Hellman, M.|author2=Raviv, J.|year=1970|title=Probability of error, equivocation, and the Chernoff bound|journal=IEEE Transactions on Information Theory|volume=16|issue=4|pp=368-372}}</ref>
<math>\Eta(Y|X)</math> is the result of averaging <math>\Eta(Y|X=x)</math> over all possible values <math>x</math> that <math>X</math> may take.
<math>\Eta(Y|X)</math> is the result of averaging <math>\Eta(Y|X=x)</math> over all possible values <math>x</math> that <math>X</math> may take.
\sum_{i=1}^n \Eta(X_i | X_1, \ldots, X_{i-1}) </math>
Sum { i = 1} ^ n Eta (x _ i | x _ 1，ldots，x _ { i-1}) </math >
Given [[Discrete random variable|discrete random variables]] <math>X</math> with image <math>\mathcal X</math> and <math>Y</math> with image <math>\mathcal Y</math>, the conditional entropy of <math>Y</math> given <math>X</math> is defined as the weighted sum of <math>\Eta(Y|X=x)</math> for each possible value of <math>x</math>, using <math>p(x)</math> as the weights:<ref name=cover1991>{{cite book|isbn=0-471-06259-6|year=1991|authorlink1=Thomas M. Cover|author1=T. Cover|author2=J. Thomas|title=Elements of Information Theory|url=https://archive.org/details/elementsofinform0000cove|url-access=registration}}</ref>{{rp|15}}
Given [[Discrete random variable|discrete random variables]] <math>X</math> with image <math>\mathcal X</math> and <math>Y</math> with image <math>\mathcal Y</math>, the conditional entropy of <math>Y</math> given <math>X</math> is defined as the weighted sum of <math>\Eta(Y|X=x)</math> for each possible value of <math>x</math>, using <math>p(x)</math> as the weights:<ref name=cover1991>{{cite book|isbn=0-471-06259-6|year=1991|authorlink1=Thomas M. Cover|author1=T. Cover|author2=J. Thomas|title=Elements of Information Theory|url=https://archive.org/details/elementsofinform0000cove|url-access=registration}}</ref>{{rp|15}}
It has a similar form to chain rule in probability theory, except that addition instead of multiplication is used.
除了用加法代替乘法之外，它的形式与概率论的链式法则相似。
:<math>
:<math>
\begin{align}
\begin{align}
Bayes' rule for conditional entropy states
条件熵的贝叶斯规则
\Eta(Y|X)\ &\equiv \sum_{x\in\mathcal X}\,p(x)\,\Eta(Y|X=x)\\
\Eta(Y|X)\ &\equiv \sum_{x\in\mathcal X}\,p(x)\,\Eta(Y|X=x)\\
\Eta(Y|X)\ &\equiv \sum_{x\in\mathcal X}\,p(x)\,\Eta(Y|X=x)\\
<math>\Eta(Y|X) \,=\, \Eta(X|Y) - \Eta(X) + \Eta(Y).</math>
[数学] Eta (y | x) ，= ，Eta (x | y)-Eta (x) + Eta (y)
& =-\sum_{x\in\mathcal X} p(x)\sum_{y\in\mathcal Y}\,p(y|x)\,\log\, p(y|x)\\
& =-\sum_{x\in\mathcal X} p(x)\sum_{y\in\mathcal Y}\,p(y|x)\,\log\, p(y|x)\\
& =-\sum_{x\in\mathcal X}\sum_{y\in\mathcal Y}\,p(x,y)\,\log\,p(y|x)\\
& =-\sum_{x\in\mathcal X}\sum_{y\in\mathcal Y}\,p(x,y)\,\log\,p(y|x)\\
Proof. <math>\Eta(Y|X) = \Eta(X,Y) - \Eta(X)</math> and <math>\Eta(X|Y) = \Eta(Y,X) - \Eta(Y)</math>. Symmetry entails <math>\Eta(X,Y) = \Eta(Y,X)</math>. Subtracting the two equations implies Bayes' rule.
证据。Eta (y | x) = Eta (x，y)-Eta (x) | math > Eta (x | y) = Eta (y，x)-Eta (y).对称意味着 Eta (x，y) = Eta (y，x)。减去这两个方程就得到了贝叶斯定律。
& =-\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log\,p(y|x)\\
& =-\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log\,p(y|x)\\
& =-\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \frac {p(x,y)} {p(x)}. \\
& =-\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \frac {p(x,y)} {p(x)}. \\
If <math>Y</math> is conditionally independent of <math>Z</math> given <math>X</math> we have:
如果[数学] y </math > 是条件独立于[数学] z </math > 给定 < 数学 > x </math > 我们有:
& = \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \frac {p(x)} {p(x,y)}. \\
& = \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \frac {p(x)} {p(x,y)}. \\
\end{align}
\end{align}
\end{align}
<math>\Eta(Y|X,Z) \,=\, \Eta(Y|X).</math>
[ math ] Eta (y | x，z) ，= ，Eta (y | x)
</math>
</math>
<!-- This paragraph is incorrect; the last line is not the KL divergence between any two distributions, since p(x) is [in general] not a valid distribution over the domains of X and Y. The last formula above is the [[Kullback-Leibler divergence]], also known as relative entropy. Relative entropy is always positive, and vanishes if and only if <math>p(x,y) = p(x)</math>. This is when knowing <math>x</math> tells us everything about <math>y</math>. ADDED: Could this comment be out of date since the KL divergence is not mentioned above? November 2014 -->
<!-- This paragraph is incorrect; the last line is not the KL divergence between any two distributions, since p(x) is [in general] not a valid distribution over the domains of X and Y. The last formula above is the [[Kullback-Leibler divergence]], also known as relative entropy. Relative entropy is always positive, and vanishes if and only if <math>p(x,y) = p(x)</math>. This is when knowing <math>x</math> tells us everything about <math>y</math>. ADDED: Could this comment be out of date since the KL divergence is not mentioned above? November 2014 -->
For any <math>X</math> and <math>Y</math>:
<！ -- 本段不正确; 最后一行不是任何两个分布之间的 KL 散度，因为 p (x)[一般]不是 x 和 y 域上的有效分布。上面的最后一个公式是 Kullback-Leibler 的背离，也被称为相对熵。相对熵总是正的，只有当且仅当数学 p (x，y) p (x) / math 时才消失。这是当我们知道数学 x / 数学告诉我们关于数学 y / 数学的一切。补充: 这个评论是否过时了，因为 KL 的分歧没有在上面提到？2014年11月--
对于任意的 < math > x </math > 和 < math > > y </math > :
<math display="block">\begin{align}
(数学显示 = “ block” > begin { align })
==Properties==
==Properties==
\Eta(Y|X) &\le \Eta(Y) \, \\
埃塔(y | x)及埃塔(y) ,
===Conditional entropy equals zero===
===Conditional entropy equals zero===
===Conditional entropy equals zero===
\Eta(X,Y) &= \Eta(X|Y) + \Eta(Y|X) + \operatorname{I}(X;Y),\qquad \\
eta (x，y) & = Eta (x | y) + Eta (y | x) + 操作数名{ i }(x; y) ，qquad
<math>\Eta(Y|X)=0</math> if and only if the value of <math>Y</math> is completely determined by the value of <math>X</math>.
<math>\Eta(Y|X)=0</math> if and only if the value of <math>Y</math> is completely determined by the value of <math>X</math>.
\Eta(X,Y) &= \Eta(X) + \Eta(Y) - \operatorname{I}(X;Y),\, \\
Eta (x，y) & = Eta (x) + Eta (y)-操作员名称{ i }(x; y) ，,
\operatorname{I}(X;Y) &\le \Eta(X),\,
操作者名{ i }(x; y) & le Eta (x) ，,
===Conditional entropy of independent random variables===
===Conditional entropy of independent random variables===
\end{align}</math>
结束{ align } </math >
Conversely, <math>\Eta(Y|X) = \Eta(Y)</math> if and only if <math>Y</math> and <math>X</math> are [[independent random variables]].
Conversely, <math>\Eta(Y|X) = \Eta(Y)</math> if and only if <math>Y</math> and <math>X</math> are [[independent random variables]].
where <math>\operatorname{I}(X;Y)</math> is the mutual information between <math>X</math> and <math>Y</math>.
其中，“数学”和“数学”之间的相互信息。
===Chain rule===
===Chain rule===
Assume that the combined system determined by two random variables <math>X</math> and <math>Y</math> has [[joint entropy]] <math>\Eta(X,Y)</math>, that is, we need <math>\Eta(X,Y)</math> bits of information on average to describe its exact state. Now if we first learn the value of <math>X</math>, we have gained <math>\Eta(X)</math> bits of information. Once <math>X</math> is known, we only need <math>\Eta(X,Y)-\Eta(X)</math> bits to describe the state of the whole system. This quantity is exactly <math>\Eta(Y|X)</math>, which gives the ''chain rule'' of conditional entropy:
Assume that the combined system determined by two random variables <math>X</math> and <math>Y</math> has [[joint entropy]] <math>\Eta(X,Y)</math>, that is, we need <math>\Eta(X,Y)</math> bits of information on average to describe its exact state. Now if we first learn the value of <math>X</math>, we have gained <math>\Eta(X)</math> bits of information. Once <math>X</math> is known, we only need <math>\Eta(X,Y)-\Eta(X)</math> bits to describe the state of the whole system. This quantity is exactly <math>\Eta(Y|X)</math>, which gives the ''chain rule'' of conditional entropy:
For independent <math>X</math> and <math>Y</math>:
对于独立的《数学》和《数学》 :
:<math>\Eta(Y|X)\, = \, \Eta(X,Y)- \Eta(X).</math><ref name=cover1991 />{{rp|17}}
:<math>\Eta(Y|X)\, = \, \Eta(X,Y)- \Eta(X).</math><ref name=cover1991 />{{rp|17}}
<math>\Eta(Y|X)\, = \, \Eta(X,Y)- \Eta(X).</math>
<math>\Eta(Y|X) = \Eta(Y) </math> and <math>\Eta(X|Y) = \Eta(X) \, </math>
Eta (y | x) = Eta (y) </math > and < math > Eta (x | y) = Eta (x) ，</math >
The chain rule follows from the above definition of conditional entropy:
The chain rule follows from the above definition of conditional entropy:
Although the specific-conditional entropy <math>\Eta(X|Y=y)</math> can be either less or greater than <math>\Eta(X)</math> for a given random variate <math>y</math> of <math>Y</math>, <math>\Eta(X|Y)</math> can never exceed <math>\Eta(X)</math>.
虽然对于给定的随机变量来说，特定条件熵的 Eta (x | y = y) </math > </math > 可能比 </math > Eta (x) </math > </math > ，< math > Eta (x | y) </math > 不能超过 math > Eta (x) </math > 。
:<math>\begin{align}
:<math>\begin{align}
\Eta(Y|X) &= \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \left(\frac{p(x)}{p(x,y)} \right) \\[4pt]
\Eta(Y|X) &= \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log \left(\frac{p(x)}{p(x,y)} \right) \\[4pt]
&= \sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)(\log (p(x))-\log (p(x,y))) \\[4pt]
The above definition is for discrete random variables. The continuous version of discrete conditional entropy is called conditional differential (or continuous) entropy. Let <math>X</math> and <math>Y</math> be a continuous random variables with a joint probability density function <math>f(x,y)</math>. The differential conditional entropy <math>h(X|Y)</math> is defined as
上面的定义适用于离散随机变量。离散条件熵的连续形式称为条件微分(或连续)熵。设 x 是连续随机变量，f (x，y)是连续随机概率密度函数。微分条件熵 < math > h (x | y) </math > 被定义为
&= -\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log (p(x,y)) + \sum_{x\in\mathcal X, y\in\mathcal Y}{p(x,y)\log(p(x))} \\[4pt]
&= -\sum_{x\in\mathcal X, y\in\mathcal Y}p(x,y)\log (p(x,y)) + \sum_{x\in\mathcal X, y\in\mathcal Y}{p(x,y)\log(p(x))} \\[4pt]
& = \Eta(X,Y) + \sum_{x \in \mathcal X} p(x)\log (p(x) ) \\[4pt]
& = \Eta(X,Y) + \sum_{x \in \mathcal X} p(x)\log (p(x) ) \\[4pt]
{{Equation box 1
{方程式方框1
& = \Eta(X,Y) - \Eta(X).
& = \Eta(X,Y) - \Eta(X).
|indent =
2012年10月22日
\end{align}</math>
\end{align}</math>
|title=
2012年10月11日
|equation = }}
| equation = }
In general, a chain rule for multiple random variables holds:
In general, a chain rule for multiple random variables holds:
|cellpadding= 6
6
|border
边界
:<math> \Eta(X_1,X_2,\ldots,X_n) =
:<math> \Eta(X_1,X_2,\ldots,X_n) =
|border colour = #0073CF
0073CF
\sum_{i=1}^n \Eta(X_i | X_1, \ldots, X_{i-1}) </math><ref name=cover1991 />{{rp|22}}
\sum_{i=1}^n \Eta(X_i | X_1, \ldots, X_{i-1}) </math><ref name=cover1991 />{{rp|22}}
|background colour=#F5FFFA}}
5/fffa }}
It has a similar form to [[Chain rule (probability)|chain rule]] in probability theory, except that addition instead of multiplication is used.
It has a similar form to [[Chain rule (probability)|chain rule]] in probability theory, except that addition instead of multiplication is used.
In contrast to the conditional entropy for discrete random variables, the conditional differential entropy may be negative.
与离散随机变量的条件熵相反，条件微分熵可能是负的。
===Bayes' rule===
===Bayes' rule===
[[Bayes' rule]] for conditional entropy states
[[Bayes' rule]] for conditional entropy states
As in the discrete case there is a chain rule for differential entropy:
在离散情况下，微分熵有一个链式规则:
:<math>\Eta(Y|X) \,=\, \Eta(X|Y) - \Eta(X) + \Eta(Y).</math>
:<math>\Eta(Y|X) \,=\, \Eta(X|Y) - \Eta(X) + \Eta(Y).</math>
<math>\Eta(Y|X) \,=\, \Eta(X|Y) - \Eta(X) + \Eta(Y).</math>
<math>h(Y|X)\,=\,h(X,Y)-h(X)</math>
H (y | x) ，= ，h (x，y)-h (x)
Notice however that this rule may not be true if the involved differential entropies do not exist or are infinite.
然而，请注意，如果所涉及的微分熵不存在或者是无限的，那么这个规则可能不成立。
''Proof.'' <math>\Eta(Y|X) = \Eta(X,Y) - \Eta(X)</math> and <math>\Eta(X|Y) = \Eta(Y,X) - \Eta(Y)</math>. Symmetry entails <math>\Eta(X,Y) = \Eta(Y,X)</math>. Subtracting the two equations implies Bayes' rule.
''Proof.'' <math>\Eta(Y|X) = \Eta(X,Y) - \Eta(X)</math> and <math>\Eta(X|Y) = \Eta(Y,X) - \Eta(Y)</math>. Symmetry entails <math>\Eta(X,Y) = \Eta(Y,X)</math>. Subtracting the two equations implies Bayes' rule.
Joint differential entropy is also used in the definition of the mutual information between continuous random variables:
联合微分熵也用于连续随机变量之间互信息的定义:
If <math>Y</math> is [[Conditional independence|conditionally independent]] of <math>Z</math> given <math>X</math> we have:
If <math>Y</math> is [[Conditional independence|conditionally independent]] of <math>Z</math> given <math>X</math> we have:
<math>\operatorname{I}(X,Y)=h(X)-h(X|Y)=h(Y)-h(Y|X)</math>
(x，y) = h (x)-h (x | y) = h (y)-h (y | x) </math >
:<math>\Eta(Y|X,Z) \,=\, \Eta(Y|X).</math>
:<math>\Eta(Y|X,Z) \,=\, \Eta(Y|X).</math>
<math>\Eta(Y|X,Z) \,=\, \Eta(Y|X).</math>
<math>h(X|Y) \le h(X)</math> with equality if and only if <math>X</math> and <math>Y</math> are independent.
当且仅当 < math > x </math > 和 < math > y </math > 是独立的。
===Other properties===
===Other properties===
For any <math>X</math> and <math>Y</math>:
For any <math>X</math> and <math>Y</math>:
For any <math>X</math> and <math>Y</math>:
The conditional differential entropy yields a lower bound on the expected squared error of an estimator. For any random variable <math>X</math>, observation <math>Y</math> and estimator <math>\widehat{X}</math> the following holds:
条件微分熵对估计量的期望平方误差产生一个下限。对于任何一个随机变量，观察值 < math > y </math > 和估计值 < math > widedhat { x } </math > ，下面是:
:<math display="block">\begin{align}
:<math display="block">\begin{align}
<math display="block">\begin{align}
<math display="block">\mathbb{E}\left[\bigl(X - \widehat{X}{(Y)}\bigr)^2\right]
< math display = " block" > mathbb { e } left [ bigl (x-widehat { x }{(y)} bigr) ^ 2 right ]
\Eta(Y|X) &\le \Eta(Y) \, \\
\Eta(Y|X) &\le \Eta(Y) \, \\
\ge \frac{1}{2\pi e}e^{2h(X|Y)}</math>
1}{2 pi e } e ^ {2 h (x | y)} </math >
\Eta(X,Y) &= \Eta(X|Y) + \Eta(Y|X) + \operatorname{I}(X;Y),\qquad \\
\Eta(X,Y) &= \Eta(X|Y) + \Eta(Y|X) + \operatorname{I}(X;Y),\qquad \\
\Eta(X,Y) &= \Eta(X) + \Eta(Y) - \operatorname{I}(X;Y),\, \\
\Eta(X,Y) &= \Eta(X) + \Eta(Y) - \operatorname{I}(X;Y),\, \\
This is related to the uncertainty principle from quantum mechanics.
这与量子力学的不确定性原理有关。
\operatorname{I}(X;Y) &\le \Eta(X),\,
\operatorname{I}(X;Y) &\le \Eta(X),\,
\end{align}</math>
\end{align}</math>
In quantum information theory, the conditional entropy is generalized to the conditional quantum entropy. The latter can take negative values, unlike its classical counterpart.
在量子信息论中，条件熵被推广为条件量子熵。后者可以采取负值，不像它的古典对应物。
where <math>\operatorname{I}(X;Y)</math> is the [[mutual information]] between <math>X</math> and <math>Y</math>.
where <math>\operatorname{I}(X;Y)</math> is the [[mutual information]] between <math>X</math> and <math>Y</math>.
For independent <math>X</math> and <math>Y</math>:
For independent <math>X</math> and <math>Y</math>:
:<math>\Eta(Y|X) = \Eta(Y) </math> and <math>\Eta(X|Y) = \Eta(X) \, </math>
:<math>\Eta(Y|X) = \Eta(Y) </math> and <math>\Eta(X|Y) = \Eta(X) \, </math>
Although the specific-conditional entropy <math>\Eta(X|Y=y)</math> can be either less or greater than <math>\Eta(X)</math> for a given [[random variate]] <math>y</math> of <math>Y</math>, <math>\Eta(X|Y)</math> can never exceed <math>\Eta(X)</math>.
Although the specific-conditional entropy <math>\Eta(X|Y=y)</math> can be either less or greater than <math>\Eta(X)</math> for a given [[random variate]] <math>y</math> of <math>Y</math>, <math>\Eta(X|Y)</math> can never exceed <math>\Eta(X)</math>.
== Conditional differential entropy ==
== Conditional differential entropy ==
=== Definition ===
=== Definition ===
The above definition is for discrete random variables. The continuous version of discrete conditional entropy is called ''conditional differential (or continuous) entropy''. Let <math>X</math> and <math>Y</math> be a continuous random variables with a [[joint probability density function]] <math>f(x,y)</math>. The differential conditional entropy <math>h(X|Y)</math> is defined as<ref name=cover1991 />{{rp|249}}
The above definition is for discrete random variables. The continuous version of discrete conditional entropy is called ''conditional differential (or continuous) entropy''. Let <math>X</math> and <math>Y</math> be a continuous random variables with a [[joint probability density function]] <math>f(x,y)</math>. The differential conditional entropy <math>h(X|Y)</math> is defined as<ref name=cover1991 />{{rp|249}}
{{Equation box 1
{{Equation box 1
Category:Entropy and information
Category:Entropy and information
类别: 熵和信息
类别: 熵和信息
|indent =
Category:Information theory
Category:Information theory