The key problem of gradient boosting is that you must maintain N decision trees to sum over, so rather than keeping only a single tree's root node in a data structure, multiple trees must remembered. The trees are summed in each inference,…

最近リスクラッテのクイズを受けながら悩んでいます。The problem is that, given a current position of y0, a position a to its left, and a position b to its right, find the probability that the particle reaches a before it reaches b. This seem…

最近オプション取引の利益分配を考えながら逆正弦法の事を味わった。 The relationship of this rule to Stirling's approximation and it's usefulness in determining the fraction of the year a trader spends in the black is unclear, but most of the…

最近確率解析の基礎を考えています。The fundamental element of the stochastic calculus seems to be the Ito integral, or the integral of a function with respect to Brownian Motion. The next important rule is Ito's Lemma, an analog of the chai…