哈工大機器學習歷年考試-資料下載頁

　　

【正文】 nny)] = Gain(S,AirTemp) = Gain(S,wind) = Gain(S,sky) =Gain(S,Humidity) = Gain(S,Forcast) = Gain(S,water) = Choose any feature of AirTemp, wind and sky as the top node.The decision tree as follow: (If choose sky as the top node)Question Four: Answer: Inductive bias: give some proor assumption for a target concept made by the learner to have a basis for classifying unseen instances.Suppose L is a machine learning algorithm and x is a set of training examples. L(xi, Dc) denotes the classification assigned to xi by L after training examples on Dc. Then the inductive bias is a minimal set of assertion B, given an arbitrary target concept C and set of training examples Dc: (Xi ) [(BDcXi) | L(xi, Dc)]C_E: the target concept is contained in the given gypothesis space H, and the training examples are all positive examples.ID3: a, small trees are preferred over larger trees. B, the trees that place high information gain attribute close to root are preferred over those that do not.BP:Smooth interpolation beteen data points.Question Five: Answer: In na239。ve bayes classification, we assump that all attributes are independent given the tatget value, while in bayes belif net, it specifes a set of conditional independence along with a set of probability distribution. Question Six： 隨即梯度下降算法 Question Seven：樸素貝葉斯例子Question Eight：The definition of three types of fitness functions in genetic algorithmAnswer: In order to select one hypothese according to fitness function, there are always three methods: roulette wheel selection, tournament selection and rank selection.Question nine:Singlepoint crossover: Crossover mask: 11111100000 or 11111000000 or 1111 0000000 or 00001111111Twopoint crossover: Offspring: (11001011000, 00101000101)Uniform crossover: Crossover mask: 10011110011 or 10001110011 or 01111101100 or 10000010011 or 10011110001 01100001100Point mutation: Any mutation is ok!1 Solution: A puter program is said to learn from experience E with respect to some class of tasks T and performance measure P,if its performance at tasks in T, as measured by P, improves with experience E.Example : (point out the T,P,E of the example)A checkers learning problem. A handwriting recognition learning problem A robot driving learning problem. ……2 Solution: S0:{, , , , , }S1:{Sunny, Warm, Normal, Strong, Warm, Same}S2:{Sunny, Warm, ?, Strong, Warm, Same}G0, G1, G2:{?, ?, ?, ?, ?, ?}S3:{ Sunny, Warm, ?, Strong, Warm, Same }G3:{Sunny, ?, ?, ?, ?, ?} {?, Warm, ?, ?, ?, ?} {?, ?, ?, ?, ?, Same}S4:{Sunny, Warm, ?, Strong, ?, ?}G4:{Sunny, ?, ?, ?, ?, ?} {?, Warm, ?, ?, ?, ?}3 Solution: (a)Here we denote S=[7+,3],then Entropy([7+,3])= =。(b)Gain(S,a2) Values()={High, Normal} ,=4 ,=5Thus Gain==4 In general,inductive inference: Some form of prior assumptions regarding the indentity of the target concept made by a learner to have a rational basis for classifying an unseen instances. Formally CANDIDATEELIMINATION:The target concept c is contained in the given hypothesis space H.Decision tree learning(ID3): Shorter trees are preferred over larger that place high information gain attributes close to the root are perferred over those that do not.BACKPROPAGATION algorithm:smooth interpolation between data points.5 Solution:(1) (2)6(3) GRADIENTDESCENT(training examples, )Each training example is a pair of the form , where is the vector of input values, and t is the target output value. is the learning rate (., ).l Initialize each to some small random valuel Until the termination condition is met, Dol Initialize each to zero.l For each in training_examples, Dol Input the instance to the unit and pute the output ol For each linear unit weight , Do a) n+1 8 歡迎您的光臨，！希望您提出您寶貴的意見，你的意見是我進步的動力。贈語； 如果我們做與不做都會有人笑，如果做不好與做得好還會有人笑，那么我們索性就做得更好，來給人笑吧！ 現在你不玩命的學，以后命玩你。我不知道年少輕狂，我只知道勝者為王。不要做金錢、權利的奴隸；應學會做“金錢、權利”的主人。什么時候離光明最近？那就是你覺得黑暗太黑的時候。最值得欣賞的風景，是自己奮斗的足跡。壓力不是有人比你努力，而是那些比你牛幾倍的人依然比你努力。 參考