This article has multiple issues. Please help improve it or discuss these issues on the talk page. (Learn how and when to remove these messages)
|
Characteristic samples is a concept in the field of grammatical inference, related to passive learning. In passive learning, an inference algorithm is given a set of pairs of strings and labels , and returns a representation that is consistent with . Characteristic samples consider the scenario when the goal is not only finding a representation consistent with , but finding a representation that recognizes a specific target language.
A characteristic sample of language is a set of pairs of the form where:
- if and only if
- if and only if
Given the characteristic sample , 's output on it is a representation , e.g. an automaton, that recognizes .
Formal Definition
editThe Learning Paradigm associated with Characteristic Samples
editThere are three entities in the learning paradigm connected to characteristic samples, the adversary, the teacher and the inference algorithm.
Given a class of languages and a class of representations for the languages , the paradigm goes as follows:
- The adversary selects a language and reports it to the teacher
- The teacher then computes a set of strings and label them correctly according to , trying to make sure that the inference algorithm will compute
- The adversary can add correctly labeled words to the set in order to confuse the inference algorithm
- The inference algorithm gets the sample and computes a representation consistent with the sample.
The goal is that when the inference algorithm receives a characteristic sample for a language , or a sample that subsumes a characteristic sample for , it will return a representation that recognizes exactly the language .
Sample
editSample is a set of pairs of the form such that
Sample consistent with a language
editWe say that a sample is consistent with language if for every pair in :
Characteristic sample
editGiven an inference algorithm and a language , a sample that is consistent with is called a characteristic sample of for if:
- 's output on is a representation that recognizes .
- For every sample that is consistent with and also fulfils , 's output on is a representation that recognizes .
A Class of languages is said to have charistaristic samples if every has a characteristic sample.
Related Theorems
editTheorem
editIf equivalence is undecidable for a class over of cardinality bigger than 1, then doesn't have characteristic samples.[1]
Proof
editGiven a class of representations such that equivalence is undecidable, for every polynomial and every , there exist two representations and of sizes bounded by , that recognize different languages but are inseparable by any string of size bounded by . Assuming this is not the case, we can decide if and are equivalent by simulating their run on all strings of size smaller than , contradicting the assumption that equivalence is undecidable.
Theorem
editIf is a characteristic sample for and is also consistent with , then every characteristic sample of , is inconsistent with .[1]
Proof
editGiven a class that has characteristic samples, let and be representations that recognize and respectively. Under the assumption that there is a characteristic sample for , that is also consistent with , we'll assume falsely that there exist a characteristic sample for , that is consistent with . By the definition of characteristic sample, the inference algorithm must return a representation which recognizes the language if given a sample that subsumes the characteristic sample itself. But for the sample , the answer of the inferring algorithm needs to recognize both and , in contradiction.
Theorem
editIf a class is polynomially learnable by example based queries, it is learnable with characteristic samples.[2]
Polynomialy characterizable classes
editRegular languages
editThe proof that DFA's are learnable using characteristic samples, relies on the fact that every regular language has a finite number of equivalence classes with respect to the right congruence relation, (where for if and only if ). Note that if , are not congruent with respect to , there exists a string such that but or vice versa, this string is called a separating suffix.[3]
Constructing a characteristic sample
editThe construction of a characteristic sample for a language by the teacher goes as follows. Firstly, by running a depth first search on a deterministic automaton recognizing , starting from its initial state, we get a suffix closed set of words, , ordered in shortlex order. From the fact above, we know that for every two states in the automaton, there exists a separating suffix that separates between every two strings that the run of on them ends in the respective states. We refer to the set of separating suffixes as . The labeled set (sample) of words the teacher gives the adversary is where is the correct lable of (whether it is in or not). We may assume that .
Constructing a deterministic automata
editGiven the sample from the adversary , the construction of the automaton by the inference algorithm starts with defining and , which are the set of prefixes and suffixes of respectively. Now the algorithm constructs a matrix where the elements of function as the rows, ordered by the shortlex order, and the elements of function as the columns, ordered by the shortlex order. Next, the cells in the matrix are filled in the following manner for prefix and suffix :
- If
- else,
Now, we say row and are distinguishable if there exists an index such that . The next stage of the inference algorithm is to construct the set of distinguishable rows in , by initializing with and iterating from the first row of downwards and doing the following for row :
- If is distinguishable from all elements in , add it to
- else, pass on it to the next row
From the way the teacher constructed the sample it passed to the adversary, we know that for every and every , the row exists in , and from the construction of , there exists a row such that and are indistinguishable. The output automaton will be defined as follows:
- The set of states is .
- The initial state is the state corresponding to row .
- The accepting states is the set .
- The transitions function will be defined , where is the element in that is indistinguishable from .
Other polynomially characterizable classes
edit- Class of languages recognizable by multiplicity automatons[4]
- Class of languages recognizable by tree automata[5]
- Class of languages recognizable by multiplicity tree automata[6]
- Class of languages recognizable by Fully-Ordered Lattice Automata[7]
- Class of languages recognizable by Visibly One-Counter Automata[8]
- Class of fully informative omega regular languages[9][10]
Non polynomially characterizable classes
editThere are some classes that do not have polynomially sized characteristic samples. For example, from the first theorem in the Related theorems segment, it has been shown that the following classes of languages do not have polynomial sized characteristic samples:
Relations to other learning paradigms
editClasses of representations that has characteristic samples relates to the following learning paradigms:
Class of semi-poly teachable languages
editA representation class is semi-poly teachable if there exist 3 polynomials , a teacher and an inference algorithm , such that for any adversary the following holds:[2]
- Selects a representation of size from
- computes a sample that is consistent with the language that recognize, of size bounded by and the strings in the sample bounded by length
- adds correctly labeled strings to the sample computed by , making the new sample of size
- then computes a representation equivalent to in time bounded by
The class of languages that there exists a polynomial algorithm that given a sample, returns a representation consistent with the sample is called consistency easy.
Polynomially characterizable languages
editGiven a representation class , and a set of identification algorithms for , is polynomially characterizable for if any has a characteristic sample of size polynomial of 's size, , that for every , 's output on is .
Releations between the paradigms
editTheorem
editA consistency-easy class has characteristic samples if and only if it is semi-poly teachable.[1]
Proof
editAssuming has characteristic samples, then for every representation , its characteristic sample holds the conditions for the sample computaed by the teacher, and the output of on every sample such that is equivalent to from the definition of characteristic sample.
Assuming that is semi-poly teachable, then for every representation , the computed sample by the teacher is a characteristic sample for .
Theorem
editIf has characteristic sample, then is polynomially characterizable.[1]
Proof
editAssuming falsely that is not polynomially characterizable, there are two non equivalent representations , with characteristic samples and respectively. From the definition of characteristic samples, any inference algorithm need to infer from the sample a representation compatible with and , in contradiction.
See also
editReferences
edit- ^ a b c d e f g h De La Higuera, Colin (1997). "[No title found]". Machine Learning. 27 (2): 125–138. doi:10.1023/A:1007353007695.
- ^ a b Goldman, Sally A.; Mathias, H.David (April 1996). "Teaching a Smarter Learner". Journal of Computer and System Sciences. 52 (2): 255–267. doi:10.1006/jcss.1996.0020. ISSN 0022-0000.
- ^ Oncina, J.; García, P. (January 1992), Inferring Regular Languages in Polynomial Updated Time, Series in Machine Perception and Artificial Intelligence, vol. 1, WORLD SCIENTIFIC, pp. 49–61, doi:10.1142/9789812797902_0004, ISBN 978-981-02-0881-3, retrieved 2024-05-21
- ^ Beimel, Amos; Bergadano, Francesco; Bshouty, Nader H.; Kushilevitz, Eyal; Varricchio, Stefano (May 2000). "Learning functions represented as multiplicity automata". Journal of the ACM. 47 (3): 506–530. doi:10.1145/337244.337257. ISSN 0004-5411.
- ^ Burago, Andrey (1994). "Learning structurally reversible context-free grammars from queries and counterexamples in polynomial time". Proceedings of the seventh annual conference on Computational learning theory - COLT '94. New York, New York, USA: ACM Press. pp. 140–146. doi:10.1145/180139.181075. ISBN 0-89791-655-7.
- ^ Habrard, Amaury; Oncina, Jose (2006), Sakakibara, Yasubumi; Kobayashi, Satoshi; Sato, Kengo; Nishino, Tetsuro (eds.), "Learning Multiplicity Tree Automata", Grammatical Inference: Algorithms and Applications, vol. 4201, Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 268–280, doi:10.1007/11872436_22, ISBN 978-3-540-45264-5, retrieved 2024-05-20
- ^ Fisman, Dana; Saadon, Sagi (2022), "Learning and Characterizing Fully-Ordered Lattice Automata", Automated Technology for Verification and Analysis, Cham: Springer International Publishing, pp. 266–282, doi:10.1007/978-3-031-19992-9_17, ISBN 978-3-031-19991-2, retrieved 2024-05-20
- ^ Berman, Piotr; Roos, Robert (October 1987). "Learning one-counter languages in polynomial time". 28th Annual Symposium on Foundations of Computer Science (SFCS 1987). IEEE. pp. 61–67. doi:10.1109/sfcs.1987.36. ISBN 0-8186-0807-2.
- ^ Angluin, Dana; Fisman, Dana (2022), Constructing Concise Characteristic Samples for Acceptors of Omega Regular Languages, arXiv:2209.09336, doi:10.1007/978-3-319-11662-4_10
- ^ Angluin, Dana; Fisman, Dana; Shoval, Yaara (2020), "Polynomial Identification of $$\omega $$-Automata", Tools and Algorithms for the Construction and Analysis of Systems, Cham: Springer International Publishing, pp. 325–343, doi:10.1007/978-3-030-45237-7_20, ISBN 978-3-030-45236-0
This article needs additional or more specific categories. (September 2024) |