-
从声速到超声波成像：A-Level 声波物理背后的完整故事
本文由理想气体的绝热过程出发，导出声波的波动方程与传播速度，推导声波的强度以及在穿过介质交界面时的强度反射系数
-
Attention Mechanism and Transformer (Deep Learning Notes C5W4)
an introduction to the transformer architecture and its underlying self-attention mechanism
-
Beam Search & BLEU Score (Deep Learning Notes C5W3)
approaches to find the optimal output and evaluate the accuracy for a machine translation task
-
从无序到正态分布：信息熵的极值原理
本篇从信息量的基本性质出发，导出信息熵的定义，并借助统计力学中的热力学熵给出其直观诠释。在此基础上，我们将证明，在均值与方差给定的约束下，使信息熵取得最大值的概率分布正是正态分布。
-
Word Embeddings (Deep Learning Notes C5W2)
generation of dense vector representations of words that can capture semantic relationships between words
-
从概率到预测：理解最大似然估计与线性回归
笔记将以正态分布为例，推导最大似然估计的求解过程，分析其统计偏差，并展示其与线性回归中误差平方和最小化之间的联系
-
LSTM and GRU Networks (Deep Learning Notes C5W1B)
introduction to LSTM and GRU networks: forward pass and backward pass
-
Recurrent Neural Networks (Deep Learning Notes C5W1A)
basics of RNNs: features and problems; forward propagation and backward pass