MODULE 1-2 MACHINE LEARNING CONCEPTSReferences Text Books: 1. Tom M. Mitchell, Machine Learning, India Edition 2013, McGraw Hill Education. Reference Books: 1, Trevor Hastie, Robert Tibshirani, Jerome Friedman, h The Elements of Statistical Learning, 2nd edition, springer series in statistics. 2, Ethem Alpaydm, Introduction to machine learning, second edition, MIT press.SYLLABUS Invoduction. Well Posed Learning Protlems, Designing a Leaning System, Perspectives and Issues in Machine Leaming, Concept Learning and the General-to Specific Ordering: Introduction, A Concept Learning Task, Concept Learning ae Search, FIND-S: Finding a Maximally Specific Hypothesis, Version Spaces and the CANDIDATE-ELIMINATION Algor, Decision Tree Learring: Ittoduction, Decision Tree Representation, Appropriate problem for Decision tro LLesrning, The Basic Decision Tree Learning Algerttnm, Hypothesis Space Search in Decision Tree Leaning, Inductive Bias in Decision Tree Leaning, lssues in Decision Tree Leaning Afiial Noural Networks: Intoduction, Natural Network Representations, Appropriate Problems for Neural Network Learning, Perceptions, Multlayer Network and the BACKPROPAGATION Algorithm ‘Bayesian Learning: Intosuction, Bayes Thecrem, Bayes Theorem and Concept Leaming, Bayes Optimal Classifier, Natve Bayes Classter,An Example: Learning to Classy Text \nstance- Based Learning: Inroduction, K-NEAREST NEIGHBOUR Learning, Distance Weighted NEAREST NEIGHBOUR Algor. Genetic Aigerithms: Motivation, Genetic Algorithms, Hypothesis Space Search, Genetic Programming, Parallaiong Genetic Algor Leaining Sets of Rules: Intoduction, Sequential Covering Algom, Learning Rule Sets: Summary Learning First-Order Rules, Learning Sets of First-Order Rules: FOIL, Induction as Inverted Deduction, ‘verted Resolution ‘Support Vector Machine: Maximum margin near separator, Quadratic Programming Solution to finding ‘maximum margin separators, Kemels fr lesmning nor-inear functions.MODULE -1Introduction Ever since computers were invented, we have wondered whether they might be made to learn. If we could understand how to program them to learn-to improve automatically with experience-the impact would be dramatic. + Imagine computers learning from medical records — which treatments are most effective for new diseases * Houses learning from experience to optimize energy costs based on the particular usage patterns of their occupants. * Personal software assistants learning the evolving interests of their users in order to highlight especially relevant stories from the online morning newspaperExamples of Successful Applications of Machine Learning + Learning to recognize spoken words + Learning to drive an autonomous vehicle + Learning to classify new astronomical structures + Learning to play world-class backgammonWhy is Machine Learning Important? = + Some tasks cannot be defined well, except by examples (e.g., recognizing people). * Relationships and correlations can be hidden within large amounts of data. Machine Learning/Data Mining may be able to find these relationships. * Human designers often produce machines that do not work as well as desired in the environments in which they are used. + The amount of knowledge available about certain tasks might be too large for explicit encoding by humans (e.g., medical diagnostic) + Environments change over time. + New knowledge about tasks is constantly being discovered by humans. It may be difficult to continuously re-design systems “by hand”.Machine Learning: A Definition + Machine learning is an application of artificial intelligence that provides systems the ability to automatically learn and improve from experience without being explicitly programmed. + Machine learning focuses on the development of computer programs that can ac data and use it learn for themselves. + A computer 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.Why “Learn”? Learning is used when: + Human expertise does not exist (navigating on Mars) + Humans are unable to explain their expertise (speech recognition) * Solution changes in time (routing on a computer network) * Solution needs to be adapted to particular cases (user biometrics)Types of Learning Supervised Learning Regression Classification Unsupervised Learning Clustering Association Reinforcement learningTypes of Learning Supervised Learning: Training these types of algorithms is having a supervisor supervising whole thing. When training a supervised learning algorithm the current training data will consist of inputs paired with correct outputs. The objective of supervised learning is to predict the correct labels for newly presented input data, Regression and classification are two types of supervised learning techniques. In case of regression output is continuous variable. Example: depending upon employee performance salary hike will be provided. Here salary hike is output variable which is a continuous value.Types of Learning Some common regression algorithms include: Linear Regression Polynomial Regression Support Vector Machine Regression Decision Tree Regression Random Forest Regression Classification: Depending upon employee performance promotion will be provided. Output is binary variable either promotion is given to the employce or it is not given. Some common classification algorithms include: Support Vector Machines Decision Trees Random Forests Naive BayeAdvantages of Supervised learning Supervised learning allows collecting data and produces data output from previous experiences. Helps to optimize performance criteria with the help of experience. Supervised machine learning helps to solve various types of real-world computation problems. It performs classification and regression tasks. It allows estimating or mapping the result to a new sample. We have complete control over choosing the number of classes we ‘want in the training data.Disadvantages of Supervised learning Classifying big data can be challenging, Training for supervised leaming needs a lot of computation time. So, it requires a lot of time. Supervised learning cannot handle all complex tasks in Machine Learning, Computation time is vast for supervised learning, It requires a labelled data set. It requires a training process.Types of Learning Supervised learning is when we teach or train the machine using data that is well-labelled. W! means some data is already tagged with the correct answer. After that, the machine is provided with a new set of examples(data) so that the supervised leaming algorithm analyses the training data(set of training examples) and produces a correct outcome from labeled data © Supervised learning involves training a machine from labeled data, © Labeled data consists of examples with the correct answer or classification. © The machine learns the relationship between inputs (images) and outputs (labels). © The trained machine can then make predictions on new, unlabeled data, © Regression: A regression problem is when the output variable is a real value, such as “dollars” or “weight”. © Classific or “blue : A classification problem is when the output variable is a category, such as “Red” , “disease” or “no disease”Unsupervised Learning ‘Unsupervised learning is a type of machine learning that learns from unlabeled data, This means that the data does not have any pre-existing labels or categories. The goal of unsupervised learning is to discover patterns and relationships in the data without any explicit guidance. ‘Unsupervised learning is the training of a machine using information that is neither classified nor labeled and allowing the algorithm to act on that information without guidance. Here the task of the machine is to group unsorted information according to similarities, pattems, and differences without any prior training of data, Unlike supervised learning, no teacher is provided that means no training will be given to the machine. Therefore the machine is restricted to find the hidden structure in unlabeled data by itself.VI i 3a nsuper ised Learning ores Key Points: © Unsupervised learning allows the model to discover patterns and relationships in unlabeled data © Clustering algorithms group similar data points together based on their inherent characteristics. © Feature extraction captures essential information from the data, enabling the model to make meaningful dis s © Label association ass characteristics 5 based on the extracted patterns andUnsupervised Learning Imagine you have a machine learning model trained on a large dataset of unlabeled images, containing both ddogs and cats, The model has never seen an image of a dog or cat before, and it has no pre-existing labels or categories for these animals. Your task is to use unsupervised learning to identify the dogs and cats in a new, unseen image. For instance, suppose it is given an image having both dogs and cats which it has never seen. Thus the machine has no idea about the features of dogs and cats so we can't categorize it as “dogs and cats But it can categorize them according to their similarities, patterns, and differences, ie, we can easily categorize the above picture into two parts. The first may contain all pics having dogs in them and the second. part may contain all pies having eats in them, Here you didn’t learn anything before, which means no ‘raining data or examples. It allows the model to work on its own to discover patterns and information that was previously undetected. It mainly deals with unlabelled dataTypes of Unsupervised Learning Unsupervised learning is classified into two categories of algorithms © Clustering: A clustering problem is where you want to discover the inherent groupings in the data, such as grouping customers by purchasing behavior. Clustering is a type of unsupervised learning that is used to group similar data points together. Clustering algorithms work by iteratively moving data points closer to their cluster centers and further away from data points in other clusters. Clustering Types: 1 2, 3 Hierarchical clustering K.means clustering Principal Component Analysis Singular Value DecompositionTypes of Unsupervised Learning Association: An association rule learning problem is where you want to discover rules that describe large portions of your data, such as people that buy X also tend to buy Y. Association rule learning is a type of unsupervised learning that is used to identify pattems in a data, Association rule leaming algorithms work by finding relationships between different items in a dataset. Some common association rule learning algorithms include: © Apriori AlgorithmAdvantages of Unsupervised learning It does not require training data to be labeled, Dimensionality reduction can be easily accomplished using unsupervised learning, Capable of finding previously unknown patterns in data Unsupervised learning can help you gain insights from unlabeled data that you might not have been able to get otherwise. Unsupervised leaming is good at finding patterns and relationships in data without being told what to look for. This can help you leam new things about your data.Application of Unsupervised learning Non-supervised learning can be used to solve a wide variety of problems, including: ‘Anomaly detection: Unsupervised learning ean identify unusual patterns or deviations from normal behavior in data, enabling the deteetion of fraud, intrusion, or system failures. Scientfie discovery: Unsupervised lesming can uncover hidden relationships and pattems in scientific data, leading to new hypotheses and insights in various scifi Feds ‘Recommendation systems: Unsupervised learning ean identity pattems and similarities in user behavior and preferences to recommend products, movies, or music that align with their interests Customer segmentation: Unsupervised learning ean identify groups of customers with similar characteristics, allowing businesses to target marketing campaigns and improve customer service more effectively Image analysis: Unsupervised learning can group images based on their content, facilitating tasks such as image classi cation, object detection, and image retrieval.Reinforcement Learning Reinforcement Learning (RL) is the science of decision making. It is about learning the optimal behavior in an environment fo obtain maximum reward, In RL, the data is accumulated from machine learning s that use a trial-and-error method. Data is not part of the input that we would find in supervised or unsupervised machine learning. Reinforcement learning uses algorithms that learn from outcomes and decide which action to take next After each action, the algorithm receives feedback that helps it determine whether the choice it made was correct, neutral or incorrect. It is a good technique to use for automated systems that have to make a lot of small decisions without human guidance, Reinforcement learning is an autonomous, self-eaching system that essentially learns by trial and error, It performs actions with the aim of maximizing rewards, or in other words, it is leaning by doing in order to achieve the best outcomesMain points in Reinforcement learning © Input: The input should be an initial state from which the model will start © Output: There are many possible outputs as there are a variety of solutions to 2 particular problem © Training: The training is based upon the input, The model will return a state and the user will decide to reward or punish the model based on its output. © The model keeps continues to learn. © The best solution is decided based on the maximum reward,Types of Reinforcement There are two types of Reinforcement 1. Positive: Positive Reinforcement is defined as when an event, occurs due to particular behavior, increases the strength and the frequency of the behavior, In other words, it has a positive effect on behavior Advantages of reinforcement learning are: © Maximizes Performance ‘© Sustain Change for a long period of time ‘© Too much Reinforcement can lead to an overload of states which can diminish the results 2. Negative: Negative Reinforcement is defined as strengthening of behavior because a negative condition is stopped or avoided. Advantages of reinforcement learning ‘© Increases Behavior ‘© Provide defiance to a minimum standard of performance ‘¢ It Only provides enough to meet up the minimum behaviorAdvantages of Reinforcement learning 1. Reinforcement learning can be used to solve very complex problems that cannot be solved by conventional techniques, 2. The model can correct the errors that occurred during the training process. 3. In RL, training data is obtained via the direct interaction of the agent with the environment 4, Reinforcement learning can handle environments that are non-deterministic, meaning that the ‘outcomes of actions are not always predictable. This is useful in real-world applications where the environment may change over time or is uncertain. 5. Reinforcement learning can be used to solve a wide range of problems, including those that involve decision making, control, and optimization, 6. Reinforcement learning is a flexible approach that can be combined with other machine learning, techniques, s leep learning, to improve performance.Disadvantages of Reinforcement learning 1. Reinforcement learning is not preferable to use for solving simple problems. 2. Reinforcement learning needs a lot of data and a lot of computation 3. Reinforcement learning is highly dependent on the quality of the reward function. If the reward function is poorly designed, the agent may not learn the desired behavior, 4, Reinforcement learning can be difficult to debug and interpret. It is not always clear why the agent is behaving in a certain way, which can make it difficult to diagnose and fix problemsWell-Posed Learning Problem Definition: A computer 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. To have a well-defined learning problem, three features needs to be identified: 1, The class of tasks 2. The measure of performance to be improved 3. The source of experienceGame Basics Checkers is played by two players. Each player begins the game with 12 colored discs. (One set of pieces is black and the other red.) Each player places his or her pieces on the 12 dark squares closest to him or her. Black moves first. Players then alternate moves. The board consists of 64 squares, alternating between 32 dark and 32 light squares. It is positioned so that each player has a light square on the right side corner closest to him or her. A player wins the game when the opponent cannot make a move. In most cases, this is because all of the opponent's pieces have been captured, but it could also be because all of his pieces are blocked in.Rules of the Game Moves are allowed only on the dark squares, so pieces always move diagonally. Single pieces are always limited to forward moves (toward the opponent). A piece making a non-capturing move (not involving a jump) may move only one square. A piece making a capturing move (a jump) leaps over one of the opponent's pieces, landing in a straight diagonal line on the other side. Only one piece may be captured in a single jump; however, multiple jumps are allowed during a single turn. When a piece is captured, it is removed from the board. Ifa player is able to make a capture, there is no option; the jump must be made. If more than one capture is available, the player is free to choose whichever he or she prefers.+ When a piece reaches the furthest row from the player who controls that piece, it is crowned and becomes a king. One of the pieces which had been captured is placed on top of the king so that it is twice as high as a single piece. Kings are limited to moving diagonally but may move both forward and backward. (Remember that single pieces, i.c. non-kings, are always limited to forward moves.) Kings may combine jumps in several directions, forward and backward, on the same turn. Single pieces may shift direction diagonally during a multiple capture turn, but must always jump forward (toward the opponent). Rules of the Game Cont.Well-Defined Learning Problem A checkers learning problem: + Task T: playing checkers + Performance measure P: percent of games won against opponents Training experience E: playing practice games against itself A handwriting recognition learning problem: + Task T: recognizing and classifying handwritten words within images + Performance measure P: percent of words correctly classified + Training experience E: a database of handwritten words with given classificationsA robot driving learning problem: + Task T: driving on public four-lane highways using vision sensors + Performance measure P: average distance travelled before an error (as judged by human overseer) + Training experience E: a sequence of images and steering commands recorded while observing a human driverDesigning a Learning System . Choosing the Training Experience .. Choosing the Target Function . Choosing a Representation for the Target Function ReNe . Choosing a Function Approximation Algorithm 1, Estimating training values 2. Adjusting the weights 5. The Final DesignCHECKERS GAME The basic design issues and approaches to machine learning is illustrated by considering designing a program to learn to play checkers, with the goal of entering it in the world checkers tournament. a . © 1. Choosing the Training Experience “= * The first design choice is to choose the type of training experience from which the system will learn. + The type of training experience available can have a__ significant impact on success or failure of the learner. There are three attributes which impact on success or failure of the leamer 1. Whether the training experience provides direct or indirect feedback regarding the choices made by the performance system. 2. The degree to which the learner controls the sequence of training examples 3. How well it represents the distribution of examples over which the final system performance P must be measured.1 regarding the choices made by the performance system. Whether the training experience provides direct or indirect feedback For example, in checkers game: In learning to play checkers, the system might learn from direct training examples consisting of individual checkers board states and the correct move for each. Indirect training examples consisting of the move sequences and final outcomes of various games played. The information about the correctness of specific moves early in the game must be inferred indirectly from the fact that the game was eventually won or lost. Here the leamer faces an additional problem of credit assignment, or determining the degree to which each move in the sequence deserves credit or blame for the final outcome. Credit assignment can be a particularly difficult problem because the game can be lost even when early moves are optimal, if these are followed later by poor moves. Hence, learning from direct training feedback is typically easier than learning from indirect feedback.2. A second important attribute of the training experience is the degree to which the learner controls the sequence of training examples For example, in checkers game: + The learner might depends on the teacher to select informative board states and to provide the correct move for each + Alternatively, the learner might itself propose board states that it finds particularly confusing and ask the teacher for the correct move. + The leamer may have complete control over both the board states and (indirect) training classifications, as it does when it leams by playing against itself with no teacher present. + Notice in this last case the learner may choose between experimenting with novel board states that it has not yet considered, or honing its skill by playing minor variations of lines of play it currently finds most promising.3. A third attribute of the training experience is how well it represents the distribution of examples over which the final system performance P must be measured. Learning is most reliable when the training examples follow a distribution similar to that of future test examples. For example, in checkers game: * In checkers learning scenario, the performance metic P is the percent of games the system wins in the world tournament. + Ifits training experience E consists only of games played against itself, there is an danger that this training experience might not be fully representative of the distribution of situations over which it will later be tested, For example, the leamer might never encounter certain crucial board states that are very likely to be played by the human checkers champion It is necessary to learn from a distribution of examples that is somewhat different from those on which the final system will be evaluated. Such situations are problematic because mastery of one distribution of examples will not necessary lead to strong performance over some other distribution,2. Choosing the Target Function The next design choice is to determine exactly what type of knowledge will be Iearned and how this will be used by the performance program. * Lets begin with a checkers-playing program that can generate the legal moves from any board state. * The program needs only to learn how to choose the best move from among these legal moves. This Icarning task is representative of a large class of tasks for which the legal moves that define some large search space are known a priori, but for which the best search strategy is not known.eS) SS i Given this setting where we must learn to choose among the legal moves, the most obvious choice for the type of information to be learned is a program, or function, that chooses the best move for any given board state. 1. Let ChooseMove _ be the target function and the notation is ChooseMove :B —>M which indicate that this function accepts as input any board from the set of legal board states B and produces as output some move from the set of legal moves M (best move). ChooseMove is an choice for the target function in checkers example, but this function will turn out to be very difficult to learn given the kind of indirect training experience available to our system2, An alternative target function is an evaluation function that assigns a numerical score to any given board state Let the target function V and the notation V:BOR which denote that V maps any legal board state from the set B to some real value. We intend for this target function V to assign higher scores to better board states. If the system can successfully learn such a target function V, then it can easily use it to select the best move from any current board position.se Let us define the target value V(b) for an arbitrary board state b in B, as follows: ae ee 1. if isa final board state that is won, then V(b) = 100 " 2. ifb isa final board state that is lost, then V(b) = -100 3. if b isa final board state that is drawn, then V(b) = 0 4, ifb isa nota final state in the game, then V(b) = V(b' ), where b! is the best final board state thatcan be achieved starting from b and playing optimally until the end of the game. Learning algorithms to acquire only some approximation to the target function so target function is often called function approximation Vv.. . fe 3. Choosing a Representation forthe “<= Target Function let us choose a simple representation - for any given board state, the function ¢ will be calculated as a linear combination of the following board features: xl: the number of black pieces on the board x2: the number of red pieces on the board x3: the number of black kings on the board x4: the number of red kings on the board x5: the number of black pieces threatened by red (i.e., which can be captured on red's next turn) x6: the number of red pieces threatened by blackThus, learning program will represent as a linear function of the form V(b) = wo + wri + worn + waxy + wars + ws + WOE Where, + w, through — w, are numerical coefficients, or weights, to be chosen by the learning algorithm. + Leamed values for the weights w, through w, will determine the relative importance of the various board features in determining the value of the board + The weight w, will provide an additive constant to the board valuePartial design of a checkers learning prograt Task T: playing checkers Performance measure P: percent of games won in the world tournament Training experience E: games played against itself Target function: V: Board —>R Target function representation V(b) = wo + wien + wox2 + w3.x3 + wary + w5x5 + WEG The first three items above correspond to the specification of the learning task, whereas the final two items constitute design choices for the implementation of the learning program.G 4. Choosing a Function Approximation “<= Algorithm * In order to learn the target function f we require a set of training examples, each describing a specific board state b and the training value V,,,,,(b) for b. (b)). + Each training example is an ordered pair of the form (b, Vi + For instance, the following training example describes a board state b in which black has won the game (note x, = 0 indicates that red has no remaining pieces) and for which the target function value V, ,,,(b) is therefore +100. ((,=3, x,=0, x,=, x,=0, x,=0, x,=0), +100)Function Approximation Procedure 1. Derive training examples from the indirect training experience available to the learner 2. Adjusts the weights w, to best fit these training examples1, Estimating training values A simple approach for estimating training values for intermediate board states is to assign the training value of V,,,,,(b) for any intermediate board state b to be vCSuccessor(b)) train| Where , Vis the learner's current approximation to V Successor(b) denotes the next board state following b for which it is again the program's turn to move Rule for estimating training values Veain(b) — ¥ (Successor(b))2. Adjusting the weights Specify the learning algorithm for choosing the weights w, to best fit the set of training examples {(b, Viai,())} A first step is to define what we mean by the best fit to the training data. + One common approach is to define the best hypothesis, or set of weights, as that which minimizes the squared error E between the training values and the values predicted by the hypothesis. E > WVerain(b) — 00)? (b, Virain(B))€ training examples + Several algorithms are known for finding weights of a linear function that minimize E.In our case, we require an algorithm that will incrementally refine the weights as new training examples become available and that will be robust to errors in these estimated training values One such algorithm is called the least mean squares, or LMS training rule. For cach observed training example it adjusts the weights a small amount in the direction that reduces the error on this training example LMS weight update rule :- For each training eg.(b,V,,,(b)) Use the current weights to calculate V(b) For each weight w,, update it as ww, — w, +n (Virain (b) - V(b)) x,(Virain(b) - ¥(@)) = 0 -> No need to change weights (Virain(6) — V(b) >0 -> Increase weight values (Virain(6) — ¥(b)) < 0 -> Decrease weight valuesHere n is a small constant (e.g., 0.1) that moderates the size of the weight update. Working of weight update rule When the error (Vtrain(b)- V(b)) is zero, no weights are changed. When (Vtrain(b) - V(b)) is positive (i.e., when V(b) is too low), then each weight is increased in proportion to the value of its corresponding feature. This will raise the value of V(b), reducing the error. If the value of some feature x; is zero, then its weight is not altered regardless of the error, so that the only weights updated are those whose features actually occur on the training example board.5. The Final Design The final design of checkers learning system can be described by four distinct program modules that represent the central components in many learning systems wey) et a) > 2 Ma >}Ce) L.The Performance System is the module that must solve the given performance “= task by using the learned target function(s). It takes an instance of a new problem (new game) as input and produces a trace of its solution (game history) as output. In checkers game, the strategy used by the Performance System to select its next move at each step is determined by the learned V evaluation function. Therefore, we expect its performance to improve as this evaluation function becomes increasingly accurate. 2.The Critic takes as input the history or trace of the game and produces as output a set of training examples of the target function. As shown in the diagram, each training example in this case corresponds to some game state in the trace, along with an estimate V__,;, of the target function value for this example.3. The Generalizer takes as input the training examples and produces an output hypothesis that is its estimate of the target function. It generalizes from the specific training examples, hypothesizing a general function that covers these examples and other cases beyond the training examples. In our example, the Generalizer corresponds to the LMS algorithm, and the output hypothesis is the function V described by the learned weights w,, ..., W, 6 4.The Experiment Generator takes as input the current hypothesis and outputs a new problem (i.c., initial board state) for the Performance System to explore. Its role is to pick new practice problems that will maximize the learning rate of the overall system. In our example, the Experiment Generator always proposes the same initial game board to begin a new game.‘The sequence of design choices made for the checkers program is summarized in below figure ‘fs nas arkOy Issues in Machine Learning Ss + What algorithms exist for learning general target functions from specific training examples? In what settings will particular algorithms converge to the desired function, given sufficient training data? Which algorithms perform best for which types of problems and representations? How much training data is sufficient? What general bounds can be found to relate the confidence in learned hypotheses to the amount of training experience and the character of the learner's hypothesis space? When and how can prior knowledge held by the learner guide the process of generalizing from examples? Can prior knowledge be helpful even when it is only approximately correct?+ What is the best strategy for choosing a useful next training experience, and how does the choice of this strategy alter the complexity of the learning problem? + What is the best way to reduce the learning task to one or more function approximation problems? Put another way, what specific functions should the system attempt to learn? Can this process itself be automated? + How can the learner automatically alter its representation to improve its ability to represent and learn the target function?MODULE-2Concept Learning + Learning involves acquiring general concepts from specific training examples. Example: People continually learn general concepts or categories such as "bird," "car," "situations in which I should study more in order to pass the exam," etc. + Each such concept can be viewed as describing some subset of objects or events defined over a larger set * Alternatively, each concept can be thought of as a Boolean-valued function defined over this larger set. (Example: A function defined over all animals, whose value is true for birds and false for other animals). Concept learning - Inferring a Boolean-valued function from training examples of its input and outputA Concept Learning Task Consider the example task of learning the target concept "Days on which my friend Aldo enjoys his favorite water sport." Strong Warm Same Yes High Strong Warm Same Yes High Strong Warm Change No High Strong Cool Change YesThe attribute EnjoySport indicates whether or not a Person enjoys his favorite water sport on this day. The task is to learn to predict the value of EnjoySport for an arbitrary day, based on the values of its other attributes ?What hypothesis representation is provided to the learner? Let’s consider a simple representation in which each hypothesis consists of a conjunction of constraints on the instance attributes. Let each hypothesis be a vector of six constraints, specifying the values of the six attributes Sky, AirTemp, Humidity, Wind, Water, and Forecast. For each attribute, the hypothesis will either + Indicate by a"? that any value is acceptable for this attribute, + Specify a single required value (¢.g., Warm) for the attribute, or + Indicate by a "O" that no value is acceptableSS If some instance x satisfies all the constraints of hypothesis h, then h classifies X as a positive example (h(x) = 1). The hypothesis that PERSON enjoys his favorite sport only on cold days with high humidity (independent of the values of the other attributes) is represented by the expression (2, Cold, High, ?, 2, ?) The most general hypothesis-that every day is a positive example-is represented by C2BWwWW? The most specific possible day is a positive _ example-is hypothesis-that no represented by (O, 0,0, 0,0, o)Notation The set of items over which the concept is defined is called the set of instances, which we denote by X. Example: X is the set of all possible days, each represented by the attributes: Sky, AirTemp, Humidity, Wind, Water, and Forecast The concept or function to be learned is called the target concept, which we denote bye. In general, c can be any boolean-valued function defined over the instances X; that is, o:X— {0, 1} Example: The target concept corresponds to the value of the attribute EnjoySport (ie., o(x) = 1 if EnjoySport = Yes, and c(x) = 0 if EnjoySport = No).SS Instances for which c(x) = 1 are called positive examples, or members of the target concept. Instances for which c(x) = 0 are called negative examples, or non-members of the target concept. The ordered pair (x, c(x)) to describe the training example consisting of the instance x and its target concept value c(x). Dto denote the set of available training examples The symbol H to denote the set of all possible hypotheses that the learner may consider regarding the identity of the target concept. Each hypothesis h in H represents a Boolean-valued function defined over X h:X—0, 1} The goal of the learner is to find a hypothesis h such that h(x) = (x) for all x in x.| = ‘Warm Wa ‘Normal High High High Strong Warm Strong Warm Strong Warm Strong Cool Same Same Change Change ‘Yes No ‘Yes oe© Given: z @ Instances X: Possible days, each described by the attributes ‘© Sky (with possible values Sunny, Cloudy, and Rainy), AirTemp (with values Warm and Cold), Humidity (with values Normal and High), Wind (with values Strong and Weak), Water (with values Warm and Cool), and Forecast (with values Same and Change). © Hypotheses H: Each hypothesis is described by a conjunction of constraints on the at- tributes Sky, AirTemp, Humidity, Wind, Water, and Forecast. The constraints may be “?” (any value is acceptable), “A” (no value is acceptable), or a specific value. Target concept c: EnjoySport : X— {0,1} @ Training examples D: Positive and negative examples of the target function (see Table 2.1). © Determine: A hypothesis h in H such that A(x) c(x) for all x in X. TABLE The EnjoySport concept learning task.3 The Inductive Learning Hypothesis Se Any hypothesis found to approximate the target function well over a sufficiently large set of training examples will also approximate the target function well over other unobserved examples.Concept learning as Search * Concept learning can be viewed as the task of searching through a large space of hypotheses implicitly defined by the hypothesis representation, + The goal of this search is to find the hypothesis that best fits the training examples. ‘Example, the instances X and hypotheses HT in the EnjoySport leaning task. The attribute Sky has three possible values, and AirTemp, Humidity, Wind, Water Forecast cach have two possible values, the instance space X contains exactly + 3.2.2.2.2.2 = 96 Distinct instances + 544.44. 5120 Syntactically distinet hypotheses within H. Every hypothesis containing one or more " " symbols represents the empty set of instances; that is, it classifies every instance as negative, + 14+ (43.33.3.3)=973. emantically distinct hypothesGeneral-to-Specific Ordering of Hypotheses + Consider the two hypotheses b= (Sunny, 2, 2 Strong, 2, ?) h,= (Sunny, 2, 2,2, 2,2) + Consider the sets of instances that are classified positive by h, and by h,. + h, imposes fewer constraints on the instance, it classifies more instances as positive. So, any instance classified positive by h, will also be classified positive by h,. Therefore, h, is more general than h,.General-to-Specific Ordering of Hypotheses > é * Given hypotheses h, and h,, h, is more-general-than or- equal do h,, if and only if any instance that satisfies h, also satisfies h, ig i Ss 1 Definition: Let h, and h, be Boolean-valued functions defined over X. Then h, is more general-than-or-equal-to h, (written h, h,) if and only if (Wx € X)[Che(x) = 1) > (4 @) = D]Instances X Hypotheses H x1= “Suny; Warm, High Ligh, Farm, Some y= Suan, 22 ® Sumy, 2,2, 2, Cook specific Genel oO In the figure, the box 2) represents the set X of all instances, the box on the right the set H of all hypotheses. Each hypothesis corresponds to some subset of X-the subset of instances that it classifies positive. The arrows connecting hypotheses represent the more - general -than relation, with the arrow pointing toward the less general hypothesis. Note the subset of instances characterized by h, subsumes the subset characterized by h , , hence h, is more - general- than h,FIND-S: Finding a Maximally Specific Hypothesis FIND-S Algorithm 1. Initialize h to the most specific hypothesis in H 2. For each positive training instance x For each attribute constraint a, in h If the constraint a, is satisfied by x Then do nothing Else replace a,in h by the next more general constraint that is satisfied by x 3. Output hypothesis /To illustrate this algorithm, assume the learner is given the sequence of training examples from the EnjoySport task Warm Normal Strong, Warm High Strong Warm Same Yes Cold High Strong Warm Change No Warm High ‘Strong Cool Change ‘Yes The first step of _FIND-S is to initialize A to the most specific hypothesis in A h-(@,@, O, D, B, B)x, = , + Observing the first training example, it is clear that our hypothesis is too specific. In particular, none of the "@" constraints in h are satisfied by this example, so each is replaced by the next more general constraint that fits the example h, = This h is still very specific; it asserts that all instances are negative except for the single positive training example x, = , + The second training example forces the algorithm to further generalize h, this time substituting a "?' in place of any attribute value in h that is not satisfied by the new example h, = x3 = , - Upon encountering the third training the algorithm makes no changeto h. The FIND-S algorithm simply ignores every negative example. h3 =< Sunny Warm ? Strong Warm Same> x4 = , + The fourth example leads to a further generalization of h hd =< Sunny Warm ? Strong ? ?>Instances X Hypotheses H SS Specific General hy= <.2,B, BB, > v= , + hy = ,+ hy = += , - hg = x= , + hy = The key property of the FIND-S algorithm is + FIND-S is guaranteed to output the most specific hypothesis within H that is consistent with the positive training examples + FIND-S algorithm final hypothesis will also be consistent with the negative examples provided the correct target concept is contained in H, and provided the training examples are correct.Example 1 Find S-Algo example | citations size inLibrary price editions | buy 1 ‘some small no affordable many | no 2 many dig no expensive one | yas 3 some big always expensive few | no 4 many medium no — expensive many | yes 5 many small no affordable many _| yes QI.How many concepts are possible for this instance space? Q2. How many hypothesis can be expressed by hypothesis language. Q3. Apply the Find S algorithm on training data given. Consider the examples in specified order and write hypothesis each time after observing an example.Example 2 Find $-Algo Example | Eyes | Nose | Head | Feolor | Hair | Smile 1 Round Triangle Round Purple ~=Yes_—=SYes 2 Square Square Square Green = Yes._—=—sNo 3 Square Triangle Round Yellow ‘Yes ‘Yes 4 Round Triangle Round Green ~=«s«No. ON 5 ‘Square Square Round Yellow ‘Yes ‘Yes QI.How many concepts are possible for this instance space? Q2. How many hypothesis can be expressed by hypothesis language. Q3. Apply the Find S algorithm on training data given, Consider the examples in specified order and write hypothesis each time after observing an example.Unanswered by FIND-S Has the learner converged to the correct target concept? Although FIND-S will find a hypothesis consistent with the training data, it has no way to determine whether it has found the only hypothesis in H consistent with the data (i.e., the correct target concept), or whether there are many other consistent hypotheses as well. We would prefer a learning algorithm that could determine whether it had converged Why prefer the most specific hypothesis? In case there are multiple hypotheses consistent with the training examples, FIND-S will find the most specific. It is unclear whether we should prefer this hypothesis over, say, the most general, or some other hypothesis of intermediate generality.Unanswered by FIND-S 3. Are the training examples consistent? In most practical learning problems there is some chance that the training examples will contain at least some errors or noise. Such inconsistent sets of training examples can severely mislead FIND-S, given the fact that it ignores negative examples. 4. What if there are several maximally specific consistent hypotheses? However, for other hypothesis spaces there can be several maximally specific hypothesis consistent with the data. In this case, FIND-S must be extended to allow it to backtrack on its choices of how to generalize the hypothesis.S Version Space and CANDIDATE ELIMINATION Algorithm The key idea in the CANDIDATE-ELIMINATION algorithm is to output a description of the set of all hypotheses consistent with the training examples Representation * Definition: A hypothesis h is consistent with a set of training examples D if and only if h(x) = (x) for each example (x, e(x)) in D. Consistent(h, D) = (W (x, e(x)) € D) h(x) = ex) Note difference between definitions of consistent and satisfies + an example x is said to satisfy hypothesis h when A(x) ~ 1, regardless of whether x is a positive or negative example of the target concept. + an example x is said to consistent with hypothesis h iff h(x) = e(x)Consistent hypothesis and version space Cece Editions | Buy 1 Some ‘Small No Affordable —One— No 2 Many aig No Expensive Many Yes ‘Assume having 2 hypothesis hl and h2. for x1; hi=<2,2,n0,?,many> Consistent h2= Not consistentVersion Space A representation of the set of all hypotheses which are consistent with D Definition: The version space, denoted VS, with respect to hypothesis space H and training examples D, is the subset of hypotheses from H consistent with the training examples in D VS, )={h € H | Consistent(h, DY}The LIST-THEN-ELIMINATE ALGORITHM The LIST-THEN-ELIMINATE algorithm first initializes the version space to contain all hypotheses in H and then climinates any hypothesis found inconsistent with any training example. Assume fl and £2 as two instances with attributes as (A,B) AND (X,Y) respectively. No of distinct instances=4 Syntactically hypothesis space=16 Semantically hypothesis=9The LIST-THEN-ELIMINATE ALGORITHM Assume training instance: FI F2 TARGET A x Yes A Y Yes Version space=(A,X)(A, Y\(A,?\(B,X)(B,Y)(B,?)(2,X)(2,Y)(2,2),0) By using LIST THEN ELIMINATION ALGORITHM Version space/ consistent hypothesis=(A,?)(?,?)The LIST-THEN-ELIMINATE ALGORITHM 1. VersionSpace ¢ a list containing every hypothesis in H 2. For each training example, (x, e(x)) remove from VersionSpace any hypothesis h for which h(x) # c(x) 3. Output the list of hypotheses in VersionSpace THE L \T-THEN-ELIMINATE PROBLEMS * List-Then-Eliminate works in principle, so long as version space is finite. + However, since it requires exhaustive enumeration of all hypotheses in practice it is not feasible.A More Compact Representation for Version Spaces * The version space is represented by its most general and least general members. * These members form general and specific boundary sets that delimit the version space within the partially ordered hypothesis space.{-) + A. version space with ‘general and specific boundary sets. Strong, 2, ?> <2, Warn, 2, Srong, 2, 2» * The version space includes all six hypotheses shown here, but can be represented more simply by S and G. + Arrows indicate instance of the a:| (, <2, Warm, 2, 2,2, 2>} ‘more-general-than relation, This is the version space for the Enjoysport concept learning, i pb ont Normal Strong Yes examples described in below table Warm = High Strong Warm = Same Yes Cold High Strong Warm Change No Warm High Strong Cool Change = Yes.Definition: The general boundary G, with _respectto_ hypothesis space H and training data D, is the set of maximally general members of H consistent with D G={g € H| Consistent(g, DNAs’ € H[(g">, 2) A Consistent(g’, D)}} Definition: The specific boundary S, with respect to hypothesis space H and training data D, is the set of minimally general (i.¢., maximally specific) members of H consistent with D. S={s €H| Consistent(s, DA(-3s' € HD[(s >,3°) A Consistent, DY]Version Space representation theorem Theorem: Let X be an arbitrary set of instances and Let H be a set of Boolean- valued hypotheses defined over X. Let ¢ : X —>{O, 1} be an arbitrary target concept defined over X, and let D be an arbitrary set of training examples {(x, c(x))). For all X, H, ¢, and D such that $ and G are well defined, VS, )={h € H |(As € S) Ag € G) (g2,h2,s)}VWSyp={h € H (As € S) Ag € G) (g2,h2,5)} To Prove: 1. Every h satisfying the right hand side of the above expression is in VS yy 2. Every member of VS ,, satisfies the right-hand side of the expression ketch of proof, 1. let g,h,s be arbitrary members of G, H, S respectively with g 2, h 2,5 By the definition of 5, s must be satisfied by all positive examples in D. Because h 2, , h must also be satisfied by all positive examples in D. By the definition of G, g cannot be satisfied by any negative example in D, and because g 2, hh cannot be satisfied by any negative example in D. Because h is satisfied by all positive examples in D and by no negative examples in D, h is consistent with D, and therefore h is a member of VS, 2.It can be proven by assuming some h in VS,, »,that does not satisfy the right-hand side of the expression, then showing that this leads to an inconsistencyThe CANDIDATE-ELIMINATION Learning Algorithm The CANDIDATE-ELIMINATION training examples. algorithm computes the version space containing all hypotheses from H that are consistent with an observed sequence of fe s > é aInitialize G to the set of maximally general hypotheses in H Initialize $ to the set of maximally specific hypotheses in H For each training example d, do + Ifd is a positive example + Remove from G any hypothesis inconsistent with d + For each hypothesis s in S that is not consistent with + Remove s from *+ Add to § all minimal generalizations h of s such that + hris consistent with d, and some member of G is more general than h + Remove from S any hypothesis that is more general than another hypothesis in S + Ifd is a negative example + Remove from S any hypothesis inconsistent with d + For each hypothesis g in G that is not consistent with d + Remove g from G *+ Add to G all minimal specializations h of g such that + his consistent with d, and some member of S is more specific than h + Remove from G any hypothesis that is less general than another hypothesis in GAn Illustrative Example The boundary sets are first initialized to G, andS,, the most general and — most specific hypotheses in H. Ss, (2, Q, @, O, ©, W))For training example d, (Sunny, Warm, Normal, Strong, Warm, Same ) + a (Sunny, Warm, Normal, Strong, Warm, Same) (2,2, 2.2.2, 2)For training example d, (Sunny, Warm, High, Strong, Warm, Same) + x (Sunny, Warm, ?, Strong, Warm, Same) (2,2, 2, 2,2.)For training example d (Rainy, Cold, High, Strong, Warm, Change )~ (Sunny, Warm, ?, Strong, Warm, Same) 2, 2)(?, Warm, ?, ?, ?, ?)(?, 2, 2, 2, 2, Same)For training example d, (Sunny, Warm, High, Strong, Cool Change ) + . (Sunny, Warm, ?, Strong, G, (Sunny, ?, ?, 2, 2, 2){?, Warm, ?, ?, ?, ?){ , <2, Warm, 2, 2, 2, 2>} The final version space for the EnjoySport concept learning problem and training examples described earlier.Example 1: Size ern Bia Crean Big Red Circle No Small Red Triangle No ‘Small Red Circle Yes Big Blue Circle No ‘Small Blue Circle Yes Compute number of consistent hypothesis for above data using candidate eliminationExample 2: Seemed Ec oe Em 1 ‘Some ‘Small No Affordable © One No 2 Many Big No Expensive Many — Yes 3 Many Medium No Expensive Few Yes. 4 Many ‘Small No Affordable © Many —Yes. Compute number of consistent hypothesis for above data using candidate climinationExample 3 : Example | Shape | Size | Color | Surface Thickness oy 1 Circular_| Large Light | Smooth | Thick Malignant (+) 2 Circular | Large | Light Irregular ‘Thick Malignant (+) 3 Oval Large -Dark =| Smooth —‘Thin Benign (-) 4 Oval | Large Light irregular. Thick Malignant (+) Compute number of consistent hypothesis for above data using candidate eliminationExample 4: Example | Eyes | Nose | Head | Feolor | Hair Smile 1 Round Triangle Round Purple Yes Yes 2 Square | Square Square Green Yes No 3 Square Triangle Round Yellow Yes Yes 4 Round Triangle Round | Green No No aes Square Square Round | Yellow Yes Yes Compute number of consistent hypothesis for above data using candidate eliminationExample 5: cy oto coor ard Gone er Type Japan Honda Blue 1980 Economy Positive Japan Toyota Green 1970 Sports Negative Japan Toyota Blue 1990 Economy Positive USA Chrysler Red 1980 Economy Negative Japan Honda White 1980 Economy _ Positive Japan Toyota Green 1980 Economy Positive Japan Honda Red 1990 Economy Negative Compute number of consistent hypothesis for above data using candidate eliminationExample Weather Temperature = WindSpend === ‘FlghtTime Delayed lar ia oe Moran Ne aim e201 ode benina ve tar Het um orang Ne Rainy ia Moderate ing Yee Snomy ald 0h renoon Yee ay Her Hen ring vee Some Ma oats srenaen Yas Compute number of consistent hypothesis for above data using candidate climinationExample 7: ‘Ase Income Gander—Education——Martalstatua Purchased 25 50000 Mle HighSehoal Single Ne m T5c00 Female Cotage Macs ve 45 60000 Mile Unisys ve a Matias _ 25 55000 Mle HighSehoot Single Ne 10000 Me Gotape singe Ne 9 72000 Fanaa Uniersty Mac vee 2950000 Mle HignSehoal Shale Ne "2 90000 Female Comage Maes vee Compute number of consistent hypothesis for above data using candidate climinationVERSION SPACE AND CANDIDATE ELIMINATION REMARKS 1. Will candidate elimination algorithm converge to correct hypothesis? 2. How can partially learned concepts be used 3. What training example should leaner request next?1. Will the CANDIDATE-ELIMINATION Algorithm Converge to the Correct Hypothesis? The version space leamed by the CANDIDATE-ELIMINATION algorithm will converge toward the hypothesis that correctly describes the target concept, provided (1) there are no errors in the training examples, and (2) there is some hypothesis in H that correctly describes the target concept. If & G converge to single hypothesis the concept is exaetly learned IfS & G do not converge to single hypothesis the concept is partially learned IfS & G is null means no hypothesis is consistent with training examples.2.How can partially learned concepts be used The learner is required to classify new instances that it has not yet observed To — <2, Warm, ?, Strong, 2, ?> NAN Gz] (,-<2, Warm, 2, 2, 22>)a Po — } > Instance Sky AirTemp — Humidity Water Forecast EnjoySport A Sunny = Warm Normal Cool Change ? B Rainy Cold-~—Normal_ Light» Warm = Same 2 C Sunny Warm = Normal Light’ Warm == Same ? D Sunny Cold. ~——Normal- Strong) Warm = Same ?3.What training example should learner request next? In this case learner should choose an instance that would be classified as positive by some of these hypothesis, but negative by others. One such instance is (Sunny, Warm, Normal, Light, Warm, Same)Inductive Bias The fundamental questions for inductive inference + What if the target concept is not contained in the hypothesis space? + Can we avoid this difficulty by using a hypothesis space that includes every possible hypothesis? + How does the size of this hypothesis space influence the ability of the algorithm to generalize to unobserved instances? + How does the size of the hypothesis space influence the number of training examples that must be observed?Effect of incomplete hypothesis space Preceding algorithms work if target function is in H Will generally not work if target function not in H Consider following examples which represent target function “sky = sunny or sky = cloudy”: {Sunny Warm Normal Strong Cool Change) (Cloudy Warm Normal Strong Cool Change) {Rainy Warm Normal Strong Cool Change) Z<< If apply Candidate Elimination algorithm as before, end up with empty Version Space After first two training example S= (2 Warm Normal Strong Cool Change) New hypothesis is overly general and it covers the third negative training example! Our H does not include the appropriate ¢An Unbiased Learner Incomplete hypothesis space + Ifc not in H, then consider generalizing representation of H to contain ¢ + The size of the instance space X of days described by the six available attributes is 96. The number of distinct subsets that can be defined over a set X containing |X| elements (ie., the size of the power set of X) is 2! + Recall that there are 96 instances in EnjoySport; hence there are 2” possible hypotheses in full space H + Can do this by using full propositional calculus with AND, OR, NOT + Hence H defined only by conjunctions of attributes is biased (containing only 973 h’ s)+ Let us reformulate the Enjoysport learning task in an unbiased way by defining a new hypothesis space H’ that can represent every subset of instances; that is, let H’ correspond to the power set of X. + One way to define such an H' is to allow arbitrary disjunctions, conjunctions, and negations of our earlier hypotheses. For instance, the target concept "Sky = Sunny or Sky = Cloudy" could then be described as (Sunny, 2, 2, 2, 2, 2) V (Cloudy, 2, 2, 2, 2,2)Definition: Consider a concept learning algorithm L for the set of instances X. + Let ¢ be an arbitrary concept defined over X + Let D,= {(x, c())} be an arbitrary set of training examples of c. + Let L(x, D,) denote the cl data De sification assigned to the instance x, by L after training on the + The inductive bias of L is any minimal set of assertions B such that for any target concept c and corresponding training examples D, (V(x, EX) [BAD.Ax) FL, DI]Induetive system, Training examples New instance Classification of Candidate new instance, or Elimination ‘don't know" Algorithm —— Hypothesis ‘quivalent deductive system Training examples New instance Assertion "Hcontains the target concept” Classification of new instance, oF ‘don't know’ Theorem Prover 7 Inductive bias made explicit Modelling inductive systems by equivalent deductive systems. The input-output behavior of the CANDIDATE-ELIMINATION algorithm using a hypothesis spa is identical to that of a deductive theorem prover utilizing the assertion ""H contains the target concept." This assertion is therefore called the inductive bias of the CANDIDATE- ELIMINATION algorithm, characterizing inductive systems by theirinductive bias allows ‘modelling them by their equivalent deductive systems. This provides a way to compare inductive systems according to theirpolicies for generalizing beyond the observed training data. Hu