A repo which record my diet habit https://diet.aaaab3n.moe
A repo which records my dietary habits
如果一个东西很难吃，那么就不会流行。
Foods with poor taste do not achieve widespread prevalence.
当一个人年龄足够大的时候，他尝过大部分流行的食物。
A sufficiently aged individual has encountered the majority of prevalent foods.
推论一: 如果一个食物你没有见过，那么它大概率是不好吃的。
Theorem. Given the First and Second Laws, the posterior probability that an unfamiliar food is bad-tasting is high.
Proof:
Let $A$ denote the event that a randomly selected food is bad-tasting, and let $B$ denote the event that the individual has not previously encountered that food.
By Bayes’ theorem:
\[P(A \mid B) = \frac{P(B \mid A) \cdot P(A)}{P(B)}\]By the First Law, bad-tasting foods are not prevalent. Combined with the Second Law, an individual has encountered most prevalent foods. Therefore, bad-tasting foods are disproportionately represented among those not yet encountered, i.e., $P(B \mid A)$ is large.
Conversely, $P(B)$ is relatively small, since by the Second Law, the set of foods not yet encountered is a small fraction of all foods.
It follows that the ratio $\frac{P(B \mid A)}{P(B)}$ is large, and thus $P(A \mid B)$ is high. $\blacksquare$
Remark (Exploration-Exploitation Tradeoff):
为了避免错失美食，应该以适当的概率 $\varepsilon$ 来尝试未出现过的食物。
To avoid converging on a local optimum and missing undiscovered palatable foods, one should adopt an $\varepsilon$-greedy strategy: with probability $\varepsilon \in [0, 1]$, explore an unfamiliar food rather than exploiting known preferences. The parameter $\varepsilon$ may be tuned adaptively via reinforcement learning methods, e.g., Monte Carlo Tree Search (MCTS).
那么大鸡排更辣的东西 cannot handle anything spicier than “Na Me Da Ji Pai” (a fried chicken cutlet brand)