Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. E. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. Information and Computation, 138(1):23--48, 10 October 1997.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....without requiring a teacher. Dean et al. 33] study the problem of learning finite automaton when the output at each state has some probability of being incorrect. They give an algorithm for learning finite automata, assuming that the robot has access to a distinguishing sequence. Freund et al. [43] give algorithms for learning typical deterministic finite automata from random walks. Deng and Papadimitriou [35] and Betke [16] model the robot s environment as a directed graph, with distinct and recognizable vertices and edges. They give a learning algorithm with a constant competitive ....

Yoav Freund, Michael Kearns, Dana Ron, Ronitt Rubinfeld, Robert Schapire, and Linda Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the Twenty-Fifth Annual ACM Symposium on Theory of Computing, pages 315--324, May 1993.

....Another approach to the examination of COTS components could utilize a simple form of behavioral specification while still relying upon execution based evaluation. We propose a methodology that combines software fault injection at component interfaces [12] and machine learning techniques [2, 7, 9, 10] in an attempt to identify problematic COTS components and understand their anomalous behavior. After a system integrator specifies certain scenarios that COTS components should not experience, our approach uses software fault injection to introduce the system to new failure situations. The usage ....

....an understanding of the anomalous behavior of certain components, our approach uses a toolkit of machine learning algorithms to model the behavior of these components. We employ techniques that learn finite automata from the behavioral data that was collected during the fault injection process [2, 7, 9]. These learning algorithms attempt to build a finite state machine where certain FSM states represent anomalous component states and specific transitions cause the component to enter these states. Refer to Section 5 for a discussion of the application of machine learning algorithms to the problem ....

[Article contains additional citation context not shown here]

Yoav Freund, Michael Kearns, Dana Ron, Ronitt Rubinfeld, Robert E. Schapire, and Linda Sellie. Efficient learning of typical finite automata from random walks. Information and Computation, 138(1):23--48, 10 October 1997.

....show that for any learning algorithm there is an adversary regular strategy for which the learning process will converge to the best response in at least exponential number of stages. One way of dealing with this complexity problem is by limiting the space of strategies available for the opponent (Freund et al. 1993; Ron and Rubinfeled, 1995) Mor et al. 1996) follow this paradigm and show that for a limited class of regular strategies, the best response automaton can be learned efficiently. In these methods, exploration is achieved by random walk, that is, incorporating randomness into the decision ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. E. Schapire, and Linda Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 25th Annual ACM Symposium on Theory and Computing, pages 315--324, 1993.

.... Pulman and Hepple, 1993; Bird, 1995; Bird and Ellison, 1994) The fact that the weaker generative capacity of FSTs makes them easier to learn than arbitrary context sensitive rules has allowed the development of a number of learning algorithms including those for deterministic finite state automata(Freund et al. 1993), deterministic transducers (Oncina, Garca, and Vidal, 1993) as well as non deterministic (stochastic) FSAs (Stolcke and Omohundro, 1993; Stolcke and Omohundro, 1994; Ron, Singer, and Tishby, 1994) Like the empiricist models we discussed above, these algorithms are all general purpose; none ....

Freund, Y., M. Kearns, D. Ron, R. Rubinfeld, R. Schapire, and L. Sellie. 1993. Efficient learning of typical finite automata from random walks. In Proc. 25rd ACM Symposium on Theory of Computing, pages 315--324.

....goal is for these labels to be the same as the labels that would be assigned by the target machine. This problem has been proven intractable in the worst case [PW 89] KV 89] In the average case, where target DFA s and training data are chosen at random, there is empirical [L 92] and theoretical [FKRRSS 93] evidence that the problem is tractable when the training data is sufficiently dense. However, when the data is sparse, the problem seems to be difficult, even in the average case. 1.2 State merging algorithms The most successful approach to date has been to use a state merging algorithm ....

Yoav Freund, Michael Kearns, Dana Ron, Ronitt Rubinfeld, Robert Schapire, and Linda Sellie. Efficient Learning of Typical Finite Automata from Random Walks, STOC-93, pp. 315-324.

....On the other hand stochastic processes are general enough to cover the sampling strategy in the PAC model as well. The first to consider PAC learning in such a setting were Aldous and Vazirani in [1] but their results do not take advantage of any properties of the function class being learnt. In [6] Freund et al. defined a specific learning model for deterministic finite automata (DFAs) which uses incremental changes. In their model a random walk follows the state graph of the automaton. It is shown that in this setting most DFAs are learnable without queries. In this paper, we consider a ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proc. 25th Annu. ACM Sympos. Theory Comput., pages 315--324. ACM Press, New York, NY, 1993.

....of a robot, have this feature. Situations like these can be modeled by stochastic processes. On the other hand stochastic processes are general enough to cover the sampling strategy in the PAC model as well. The first to consider PAC learning in such a setting were Aldous and Vazirani in [1] In [8] Freund et al. defined a specific learning model for DFAs which essentially uses incremental changes. In their model a random walk follows the state graph of the automaton. It is shown that in this setting most DFAs are learnable without queries. In this paper, we consider a general model where ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proc. 25th Annu. ACM Sympos. Theory Comput., pages 315--324. ACM Press, New York, NY, 1993.

....consisting of all strings out to a given length. 6] showed empirically that this same algorithm can often construct an approximately correct hypothesis from a sparse subset of a complete training set, when both the target concept and training sets are randomly chosen from uniform distributions. [8] proved the approximate learnability of DFA s with worst case graph structure and randomly labeled states, from randomly chosen training strings. 1 We note that many papers have been published on the application of generic methods such as neural networks and genetic search to the problem of DFA ....

....training set construction, so it was eliminated by selecting only those graphs with a depth 2 of exactly 2 log 2 n 2. A training set for a target of nominal size n consisted of a random sample drawn without replacement from a uniform distribution over the collection of 16n 2 1 binary 1 The [8] theorem concerns a slightly different protocol, in which the learner sees the label of every state that is encountered rather than just the label of the final state. 2 By analogy to trees, the depth of a DFA is maximum over all nodes x of the length of the shortest path from the root to x. ....

Yoav Freund, Michael Kearns, Dana Ron, Ronitt Rubinfeld, Robert Schapire, and Linda Sellie. Efficient Learning of Typical Finite Automata from Random Walks, STOC-93, pp. 315-324.

No context found.

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. E. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. Information and Computation, 138(1):23--48, 10 October 1997.

....regular expressions, then the problem reduces to learning a regular set. Unfortunately, it has been shown that learning a regular set is inefficient under widely held 3 assumptions [5] There are several promising heuristic approaches, and it is currently an area of considerable active research [3]. In addition, our problem is made more difficult than the problem of learning a regular set, because we are interested in formulating regular sets from subsequences of a program trace. Thus, we do not have positive and negative examples for automatic learning, as would normally be the case. 3 ....

Yoav Freund, Michael Kearns, Dana Ron, Ronitt Rubinfeld, Robert E. Schapire, and Linda Sellie. Efficient learning of typical finite automata from random walks. ACM STOC, 1993. 9

No context found.

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R.E. Schapire, and L. Sellie, Efficient Learning of Typical Finite Automata from Random Walks, STOC-93.

....[2] describes an algorithm for learning DFAs given access both to random examples and to membership queries. Rivest and Schapire [30] 31] 33] present algorithms for inferring DFAs from input output behavior in the absence of a means of resetting the machine to a start state. Freund et al. [12] present efficient algorithms for learning typical DFAs from random walks without membership queries, both when the learner is provided with the means of resetting the machine and when it is not. Several results have been obtained for learning in the presence of errors in the Probably ....

Y. Freund, M. J. Kearns, D. Ron, R. Rubinfeld, R. E. Schapire, and L. Sellie. (1993). Efficient learning of typical finite automata from random walks. Proceedings of the 25th Annual ACM Symposium on Theory of Computing (pp. 315--324). San-Diego, CA: The Association for Computing Machinery.

....is also polynomial in a parameter related to the mixing rate (or the correlation length) of the target machine. Hoffgen [6] studies related families of distributions, but his algorithms depend exponentially and not polynomially on the order, or memory length, of the distributions. Freund et al. [5] point out that their result for learning typical deterministic finite automata from random walks without membership queries, can be extended to learning typical PFAs. Unfortunately, there is strong evidence indicating that the problem of learning general PFAs is hard. Kearns et al. 8] show ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R.E. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 24th Annual ACM Symposium on Theory of Computing, pages 315--324, 1993.

....uses automata which have hard to reach states and hence exponential cover time. In fact, Angluin shows that if you have a method of efficiently reaching all states of the automaton (which is true for graphs with polynomial cover time) then you can exactly learn the automaton using reset. In [7], it is shown how to learn typical automata (automata in which the underlying graph is arbitrary, but the accept reject labels on the states are chosen randomly) by passive learning (the edge traversed by the robot is chosen randomly) In [12] an algorithm is given for learning DFAs with reset ....

Yoav Freund, Michael J. Kearns, Dana Ron, Ronitt Rubinfeld, Robert E. Schapire, and Linda Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 24th Annual ACM Symposium on Theory of Computing, 1993.

....dependencies in the sequences, then these models are clearly not practical. Hoffgen [7] studies families of distributions related to the ones studied in this paper, but his algorithms depend exponentially and not polynomially on the order, or memory length, of the distributions. Freund et al. [5] point out that their result for learning typical deterministic finite automata from random walks without membership queries, can be extended to learning typical PFAs. Unfortunately, there is strong evidence indicating that the problem of learning general PFAs is hard. Kearns et al. 11] show ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R.E. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 24th Annual ACM Symposium on Theory of Computing, pages 315--324, 1993.

....form of signatures of states in order to distinguish between the states of the target automaton. This technique was presented in the pioneering work of Trakhtenbrot and Brazdin [20] in the context of learning deterministic finite automata (DFAs) The same idea was later applied by Freund et al. [6] in their work on learning typical DFAs 1 . In the same work they proposed to apply the notion of statistical signatures 1 They define typical DFAs to be DFAs in which the underlying graph is arbitrary, but the accept reject labels on the states are chosen randomly. to learning typical PFAs. ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. E. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 24th Annual ACM Symp. on Theory of Computing, pages 315--324, 1993.

....actually learns the topology of the underlying graph. Their algorithms (with the exception of one, for permutation automata) rely on a teacher. The teacher supplies counterexamples to the robot s hypotheses. Variants of this problem that do not rely on a teacher are studied in the following works [14, 18, 28, 17]. We note that Dean et al. 14] apply a cycling technique related to ours, but for different purposes. Exploring and navigating in geometric environments is studied extensively. A sample of papers includes [5, 23, 15, 11, 6, 10, 8, 19, 2] 2 Preliminaries Let G = V;E) be the unknown directed ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 24th Annual ACM Symposium on Theory of Computing, pages 315--324, 1993.

....form of signatures of states in order to distinguish between the states of the target automaton. This technique was presented in the pioneering work of Trakhtenbrot and Brazdin [21] in the context of learning deterministic finite automata (DFAs) The same idea was later applied by Freund et al. [6] in their work on learning typical DFAs 1 . In the same work they proposed to apply the notion of statistical signatures to learning typical PFAs. The outline of our learning algorithm is roughly the following. In the course of the algorithm 1 They define typical DFAs to be DFAs in which the ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R.E. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 24th Annual ACM Symp. on Theory of Computing, pages 315--324, 1993.

....actually learns the topology of the underlying graph. Their algorithms (with the exception of one, for permutation automata) rely on a teacher. The teacher supplies counterexamples to the robot s hypotheses. Variants of this problem that do not rely on a teacher are studied in the following works [14, 18, 27, 17]. We note that Dean et al. 14] apply a cycling technique related to ours for different purposes. Exploring and navigating in geometric environments is studied extensively. A sample of papers includes [5, 22, 15, 11, 6, 10, 8, 19, 2] 2 Preliminaries Let G = V; E) be the unknown directed graph ....

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 24th Annual ACM Symposium on Theory of Computing, pages 315--324, 1993.

No context found.

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R.E. Schapire and L. Sellie. "Efficient learning of typical finite automata from random walks", 25th ACM Symposium on the Theory of Computing (1993), 315-324.

No context found.

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R.E. Schapire and L. Sellie. "Efficient learning of typical finite automata from random walks", 25th ACM Symposium on the Theory of Computing (1993), 315--324.

No context found.

Y. Freund et al., "Efficient Learning of Typical Finite Automata from Random Walks", 25th ACM Symposium on Theory of Computing , pp.315324, 1993.

No context found.

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. Schapire, and L. Sellie (1993) Efficient Learning of Typical Finite Automata from Random Walks, STOC-93, pp. 315-324.

No context found.

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In Proceedings of the 25th Annual ACM Symposium on the Theory of Computing, pages 315--324, 1993.

No context found.

Y. Freund, M. Kearns, D. Ron, R. Rubinfeld, R. E. Schapire, and L. Sellie. Efficient learning of typical finite automata from random walks. In ACM Symposium on the Theory of Computing, pages 315--324, 1993.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC