数学¶
addition: 加
subtraction: 减
multiplication: 乘
product: 乘积
differentiable: 可微 * differentiability: 可微性 * 反向传播从最终损失值开始,从最顶层反向作用至最底层,利用链式法则计算每个参数对损失值的贡献大小。 * Backpropagation starts with the final loss value and works backward from the top layers to the bottom layers, applying the chain rule to compute the contribution that each parameter had in the loss value.
derivate: n. 导数;派生词;派生的事物 * derivative: n. [数] 导数 [化学] 衍生物;金融衍生产品;派生词; * Partial derivative: 偏导数
Linear Algebra[ˈældʒɪbrə]: 线性代数
Probability Statistics: 概率统计
probability: 概率论
Calculus: 微积分学
deviation: 偏差 * Standard deviation: 标准偏差
Covariance: 协方差
Entropy: 熵
Multivariable Calculus: 多元微积分
squared error: 平方误差
Geometric: 几何
scalar: n. [数] 标量, adj. 标量的;数量的;梯状的,分等级的
vector: n. [数] 向量;矢量;带菌者;航线, vt. 用无线电导航
matrix: n. [数] 矩阵;模型;社会环境;基质;母体;子宫;脉石
tensor: n. [数] 张量;[解剖] 张肌
logarithm: n. [数] 对数
exponential: n. [数] 指数
Gradient: n. [数] 梯度;坡度;倾斜度 * stochastic gradient descent(SGD) * mini-batch SGD: 小批量随机梯度下降 * true SGD: 真SGD * batch SGD: 批量 SGD
Hadamard: 阿达玛(Hadamard)矩阵:简称H矩阵. ®H矩阵是一个方阵,仅有元素+1和-1构成,而且其各行(和列) 是互相正交的
identity matrix: n. [数] 单位矩阵
the indicator function: 指标函数
transpose of a vector or a matrix: 一个向量或矩阵的转置
Inverse of matrix X: 矩阵X的逆矩阵
Gradient of 𝑦 with respect to 𝐱: y关于x的梯度
Definite integral of 𝑓 from 𝑎 to 𝑏 with respect to 𝑥: f关于x从a到b的定积分
Random variable 𝑧 has probability distribution 𝑃: 随机变量z的概率分布为p
loss function: 损失函数
clustering algorithm: 聚类算法
Standard deviation: 标准偏差
normal equation: 正规方程
Probability distribution: 概率分布
Bernoulli distribution: 伯努利分布
Gaussian distribution: 高斯分布
Monte Carlo tree: 蒙特卡罗树
Conditional probability: 条件概率
Probability density function: [数] 机率密度函数;概率密度分布函数
linear regression: 线性回归
logistic regression: 逻辑回归
stochastic gradient descent: 随机梯度下降
multi-layer perceptrons: 多层感知器
pixel /’pɪks(ə)l/ /’pɪksl/: 像素,分辨率
the 𝐿2 loss corresponds to the assumption that our data was corrupted by Gaussian noise, whereas the 𝐿1 loss corresponds to an assumption of noise from a Laplace distribution.
Bayesian analysis: 贝叶斯分析