梁启超的古籍辨伪方法    梁启超先生不仅是近代史界革命的倡导者，也是古籍辨伪的宣传者与实践者，同时还是传统辨伪方法的总结者。1922年，梁启超先生正式发表了《中国历史研究法》，1927年发表《古书真伪及其年代》，对古籍辨伪的重要性与方法进行了论述。     什麽是伪书？伪书有哪几种情况？梁启超先生在《古书真伪及其年代》中「伪书的种类及作伪的来历」一章中归纳了十种情况：     一、为全部伪；     二、一部伪；     三、本无其书而伪；     四、曾有其书，因佚而伪；     五、内容不尽伪，而书名伪；     六、内容不尽伪，而书名人名皆伪；     七、内容及书名不伪而人名伪；     八、盗袭割裂旧书而伪；     九、伪后出伪；     十、伪中益伪。     换言之，凡有以上十类问题的图书皆可入「伪书」之列。按照这个标准，在传世先秦两汉文献中，不是「伪书」的大概就没有几种了，就连《论语》、《史记》等等都包括在内，因此梁先生也不禁感慨道：「中国的伪书，真是多极了！」[7] 显然，梁启超先生提出的上述伪书的标准，有些是有问题的。     在辨伪方法上，梁启超先生在总结前人特别是胡应麟辨伪方法的基础上，将古籍辨伪的方法归纳为两方面，即：从传授统绪上和文义内容上辨别。如何从传授统绪上辨别？梁启超先生提出了八种办法：     一、从旧志不着录，而定其伪或可疑；     二、从前志着录，后志已佚，而定其伪或可疑；     三、从今本和旧志说的卷数篇数不同，而定其伪或可疑；     四、从旧志无着者姓名，而定后人随便附上去的姓名是伪；     五、从旧志或注家已明言是伪书而信其说；     六、后人说某书出现于某时，而那时人并未看见那书，从这上可断定那书是伪；     七、书初出现，已发生许多问题，或有人证明是伪造，我们当然不能相信；     八、从书的来历暧昧不明而定其伪。     如何从文义内容上来辨别？梁启超先生提出了「五大法门」：     一、从字句缺漏处辨别（如从人的称谓上辨别、从用后代的人名地名朝代名来辨别、从后代的事实或法制来辨别等等）；     二、从抄袭旧文处辨别；     三、从佚文上辨别；     四、从文章上辨别（如从名词、文体、文法、音韵等等）；     五、从思想上辨别（如从思想系统和传授家法辨别、从思想和时代的关係辨别、从专门术语和思想的关係辨别、从袭用后代学说辨别。[8]     胡适先生在《中国哲学史大纲》中也对古籍的辨伪工作给予了高度的重视，尤其对梁启超先生从文义内容上辨别伪书的「五大法门」极为推崇。除在《古书真伪及其年代》中对辨伪之法进行了专门论述，梁启超先生在《中国历史研究法》中还提出了辨伪书的十二条「公例」。     较之胡应麟的辨伪八法，梁启超先生的辨伪方法无疑要全面得多，因此备受学术界的重视，并且在古籍的考辨中被广泛应用。

Back prop is just gradient descent on individual errors. You compare the predictions of the neural network with the desired output and then compute the gradient of the errors with respect to the weights of the neural network. This gives you a direction in the parameter weight space in which the error would become smaller.Interestingly, due to the layered structure of the neural network and the chain rule for derivatives, the formulas you get can be interpreted as propagating the error back through the network. But that's mostly a computational aside, what you really (just) do is gradient descent, that is, changing the weights of the neural network a little bit to make the error on your training examples smaller.

反向传播只是在个别错误上进行梯度下降。通过比较对神经网络预期输出的预测，而后计算相对于神经网络的权重的误差梯度。然后得出了权值空间中减小误差的方向。

为了采用反向传播（见下文），神经网络必须使用可微的激活函数。

Every new idea is just a mashup or a remix of one or more previous ideas

BP算法的思想是：将训练误差E看作以权重向量每个元素为变量的高维函数，通过不断更新权重，寻找训练误差的最低点，按误差函数梯度下降的方向更新权值。

激活函数在神经网络中的功能即通过对加权的输入进行非线性组合产生非线性决策边界

Distributional Hypothesis, which states that words that appear in the same contexts share semantic meaning

每层的数学理解：用线性变换跟随着非线性变化，将输入空间投向另一个空间。

Topology of tanh Layers Each layer stretches and squishes space, but it never cuts, breaks, or folds it. Intuitively, we can see that it preserves topological properties. For example, a set will be connected afterwards if it was before (and vice versa). Transformations like this, which don’t affect topology, are called homeomorphisms. Formally, they are bijections that are continuous functions both ways. Theorem: Layers with NNN inputs and NNN outputs are homeomorphisms, if the weight matrix, WWW, is non-singular. (Though one needs to be careful about domain and range.) Proof: Let’s consider this step by step: Let’s assume WWW has a non-zero determinant. Then it is a bijective linear function with a linear inverse. Linear functions are continuous. So, multiplying by WWW is a homeomorphism. Translations are homeomorphisms tanh (and sigmoid and softplus but not ReLU) are continuous functions with continuous inverses. They are bijections if we are careful about the domain and range we consider. Applying them pointwise is a homemorphism Thus, if WWW has a non-zero determinant, our layer is a homeomorphism. ∎ This result continues to hold if we compose arbitrarily many of these layers together.

CBOW: The input to the model could be wi−2,wi−1,wi+1,wi+2wi−2,wi−1,wi+1,wi+2w_{i-2}, w_{i-1}, w_{i+1}, w_{i+2}, the preceding and following words of the current word we are at. The output of the neural network will be wiwiw_i. Hence you can think of the task as "predicting the word given its context"Note that the number of words we use depends on your setting for the window size.Skip-gram: The input to the model is wiwiw_i, and the output could be wi−1,wi−2,wi+1,wi+2wi−1,wi−2,wi+1,wi+2w_{i-1}, w_{i-2}, w_{i+1}, w_{i+2}. So the task here is "predicting the context given a word". Also, the context is not limited to its immediate context, training instances can be created by skipping a constant number of words in its context, so for example, wi−3,wi−4,wi+3,wi+4wi−3,wi−4,wi+3,wi+4w_{i-3}, w_{i-4}, w_{i+3}, w_{i+4}, hence the name skip-gram.

"It's kind of surprising because chaos is this totally classical concept—there's no idea of chaos in a quantum system," Charles Neill, a researcher in the UCSB Department of Physics and lead author of a paper that appears in Nature Physics. "Similarly, there's no concept of entanglement within classical systems. And yet it turns out that chaos and entanglement are really very strongly and clearly related."

organizations which design systems ... are constrained to produce designs which are copies of the communication structures of these organizations — M. Conway

What is the adaptive value of consciousness?

deliberate practice

一部书的奇迹：虽然1997年去世的杰恩斯从来没有完成过其他书，但是，这本书将会让他名留千古。约翰·厄普代克在《纽约客》中写道：杰恩斯「推论直到公元前2000年人类都没有意识，而是自发地遵从上帝的声音，我们都震惊了。但是，通过了解所有他在古代文学、现代行为主义和诸如催眠、着魔、舌语现象、预言、诗歌和精神分裂症等异常的心理现象中找到的确凿证据，我们不得不接受这一非凡的论题。」

When Bodhisattva Avalokitesvara practised the deep Prajnaparamita, he saw that the five skandhas were empty; thus he overcame all ills and suffering. "O Sariputra! Form does not differ from the void, and the void does not differ from the form. Form is the void, and the void is form. The same is true for feelings, conceptions, impulses and consciousness. O Sariputra, the characteristics of the void is not created, not annihilated, not impure, not pure, not increasing, not decreasing. Therefore, in the void there are no forms and no feelings, conceptions, impulses and no consciousness: there is no eye, ear, nose, tongue, body or mind; there is no form, sound, smell, taste, touch or idea; no eye elements, until we come to no elements of consciousness; no ignorance and also no ending of ignorance, until we come to no old age and death; and no ending of old age and death. Also, there is no truth of suffering, of the cause of suffering, of the cessation of suffering or of the path. There is no wisdom, and there is no attainment whatsoever. Because there is nothing to be attained, a Bodhisattva relying on Prajnaparamita has no obstruction in his heart. Because there is no obstruction he has no fear, and he passes far beyond all confused imagination and reaches Ultimate Nirvana. All Buddhas in the past, present and future have attained Supreme Enlightenment by relying on the Prajnaparamita. Therefore we know that the Prajnaparamita is the great magic Mantra, the great Mantra of illumination, it is the supreme Mantra, the unequaled Mantra which can truly wipe out all suffering without fail." Therefore, he uttered the Prajnaparamita mantra, by saying: "Gate, Gate, Paragate, Parasemgate Bodhi-svaha!"

Form is emptiness, Emptiness is form

当九十年代末，予友李公洪巖、范公旭仑考证钱锺书留学英国，决不可能获副博士之苏联学位，又谓杨绛《钱锺书与〈围城〉》非信史，杨便致电丁关根同志哭诉，请将李、范书销禁。令遂行。而中书君身后，杨绛终不得不推翻昨日之我，承认钱所戴者为学士帽，而非苏联友好援助英联邦之副博士帽也。 棣之师尝谓，钱锺书心中孙柔嘉必系杨绛无疑，而必语之云：汝卽唐晓芙了。唉，在敌人眼皮底下幹革命，用句京戏词儿说：忒亦地辛苦了！

The hermeneutic circle (German: hermeneutischer Zirkel) describes the process of understanding a text hermeneutically. It refers to the idea that one's understanding of the text as a whole is established by reference to the individual parts and one's understanding of each individual part by reference to the whole. Neither the whole text nor any individual part can be understood without reference to one another, and hence, it is a circle. However, this circular character of interpretation does not make it impossible to interpret a text; rather, it stresses that the meaning of a text must be found within its cultural, historical, and literary context.

After graduating from MIT at the age of 29, Loveman began teaching at Harvard Business School, where he was a professor for nine years.[8][10] While at Harvard, Loveman taught Service Management and developed an interest in the service industry and customer service.[8][10] He also launched a side career as a speaker and consultant after a 1994 paper he co-authored, titled "Putting the Service-Profit Chain to Work", attracted the attention of companies including Disney, McDonald's and American Airlines. The paper focused on the relationship between company profits and customer loyalty, and the importance of rewarding employees who interact with customers.[7][8] In 1997, Loveman sent a letter to Phil Satre, the then-chief executive officer of Harrah's Entertainment, in which he offered advice for growing the company.[7] Loveman, who had done some consulting work for the company in 1991,[11] again began to consult for Harrah's and, in 1998, was offered the position of chief operating officer.[8] He initially took a two year sabbatical from Harvard to take on the role of COO of Harrah's,[10] at the end of which Loveman decided to remain with the company.[12]

the most important figures that one needs for management are unknown or unknowable (Lloyd S. Nelson, director of statistical methods for the Nashua corporation), but successful management must nevertheless take account of them.

The structure of history, Ranke argues, echoes the structure of a text in so far as it consists of a particular kind of co-dependence between parts and whole. Like reading, understanding history means moving along the paths of the hermeneutic circle, from part to whole and back again. Because the historical mind is itself situated in history, there is, however, no end to this circular movement.