Showing papers on "Sequence learning published in 1994"

Motor sequence learning: a study with positron emission tomography

Abstract: We have used positron emission tomography to study the functional anatomy of motor sequence learning. Subjects learned sequences of keypresses by trial and error using auditory feedback. They were scanned with eyes closed under three conditions: at rest, while performing a sequence that was practiced before scanning until overlearned, and while learning new sequences at the same rate of performance. Compared with rest, both sequence tasks activated the contralateral sensorimotor cortex to the same extent. Comparing new learning with performance of the prelearned sequence, differences in activation were identified in other areas. (1) Prefrontal cortex was only activated during new sequence learning. (2) Lateral premotor cortex was significantly more activated during new learning, whereas the supplementary motor area was more activated during performance of the prelearned sequence. (3) Activation of parietal association cortex was present during both motor tasks, but was significantly greater during new learning. (4) The putamen was equally activated by both conditions. (5) The cerebellum was activated by both conditions, but the activation was more extensive and greater in degree during new learning. There was an extensive decrease in the activity of prestriate cortex, inferotemporal cortex, and the hippocampus in both active conditions, when compared with rest. These decreases were significantly greater during new learning. We draw three main conclusions. (1) The cerebellum is involved in the process by which motor tasks become automatic, whereas the putamen is equally activated by sequence learning and retrieval, and may play a similar role in both. (2) When subjects learn new sequences of motor actions, prefrontal cortex is activated. This may reflect the need to generate new responses. (3) Reduced activity of areas concerned with visual processing, particularly during new learning, suggests that selective attention may involve depressing the activity of cells in modalities that are not engaged by the task.

Characteristics of dissociable human learning systems

Abstract: A number of ways of taxonomizing human learning have ben proposed. We examine the evidence for one such proposal, namely, that there exist independent explicit and implicit learning systems. This combines two further distinctions, (1) between learning that takes place with versus without concurrent awareness, and (2) between learning that involves the encoding of instances (or fragments) versus the induction of abstract rules or hypotheses. Implicit learning is assumed to involve unconscious rule learning. We examine the evidence for implicit learning derived from subliminal learning, conditioning, artificial grammar learning, instrumental learning, and reaction times in sequence learning. We conclude that unconscious learning has not been satisfactorily established in any of these areas. The assumption that learning in some of theses tasks (e.g., artificial grammar learning) is predominantly based on rule abstraction is questionable. When subjects cannot report the ''implicitly learned'' rules that govern stimulus selection, this is often because their knowledge consists of instances or fragments of the training stimuli rather than rules. In contrast to the distinction between conscious and unconscious learning, the distinction between instance and rule learning is a sound and meaningful way of taxonomizing human learning. We discuss various computational models of these two forms of learning.

Learning, Remembering, Believing. Enhancing Human Performance.

Abstract: Can such techniques as sleep learning and hypnosis improve performance? Do we sometimes confuse familiarity with mastery? Can we learn without making mistakes? These questions apply in the classroom, in the military, and on the assembly line. Learning, Remembering, Believing addresses these and other key issues in learning and performance. The volume presents leading-edge theories and findings from a wide range of research settings: from pilots learning to fly to children learning about physics by throwing beanbags. Common folklore is explored, and promising research directions are identified. The authors also continue themes from their first two volumes: Enhancing Human Performance (1988) and In the Mind's Eye (1991). The result is a thorough and readable review of * Learning and remembering. The volume evaluates the effects of subjective experience on learning--why we often overestimate what we know, why we may not need a close match between training settings and real-world tasks, and why we experience such phenomena as illusory remembering and unconscious plagiarism. * Learning and performing in teams. The authors discuss cooperative learning in different age groups and contexts. Current views on team performance are presented, including how team-learning processes can be improved and whether team-building interventions are effective. * Mental and emotional states. This is a critical review of the evidence that learning is affected by state of mind. Topics include hypnosis, meditation, sleep learning, restricted environmental stimulation, and self-confidence and the self-efficacy theory of learning. * New directions. The volume looks at two new ideas for improving performance: emotions induced by another person--socially induced affect--and strategies for controlling one's thoughts. The committee also considers factors inherent in organizations--workplaces, educational facilities, and the military--that affect whether and how they implement training programs. Learning, Remembering, Believing offers an understanding of human learning that will be useful to training specialists, psychologists, educators, managers, and individuals interested in all dimensions of human performance.

Reinforcement-learning control and pattern recognition systems

Abstract: Publisher Summary This chapter presents the formulation of the learning models and its application to the synthesis of control systems and pattern-recognition systems in such a manner that these systems may be said to possess attributes of learning. Learning may be defined as the process by which an activity originates or is changed through reaction to an encountered situation which is not due to “native” response tendencies, maturation, or temporary states of the organism. The chapter also discusses the basic properties of reinforcement-learning pattern recognition systems. There are two types of control systems—open-loop and closed-loop. Just as for learning control systems, correspondences exist between elements of stochastic learning theory and pattern recognition systems. Mathematical learning models combine the mathematical properties and psychological concepts of learning. They are formulated from experimental results and attempt to predict learning behavior quantitatively. One of the most important principles in all learning theory is the law of reinforcement.

The influence of progress tests and block tests on study behaviour

Abstract: In a problem-based learning system the acquisition of knowledge is initiated and structured by the analysis of problem tasks. Students are expected to state learning objectives for themselves, depending on their interpretation of the problem description. It is claimed that this process of self responsible, intrinsically motivated learning results in stable, firmly anchored knowledge and establishes the students' steady growth in the professional domain. This could be contrasted with the short term oriented, test directed learning activities in traditional curricula, where students' mastery of a (sub)domain of knowledge is, to a large degree, course bounded. In the construction of assessment procedures within a problem-based curriculum special consideration has to be given to the congruence of instruments with the claims as mentioned regarding student behaviour. Progress testing is used as a means for leaving the maximum freedom of learning to students whilst still validly cover the content of the professional domain. Besides this category of knowledge testing other tests play a more or less dominant role, such as testing of professional skills, and, to some degree, knowledge testing in connection with specific courses (so called block tests). In this study the role of progress testing in students' behaviour and students' perceptions are explored within the context of the Health Sciences curriculum in Maastricht. The findings support the claim that progress testing favors learning processes that are characterized as meaning oriented in contrast to reproduction oriented learning. Success in block tests, on the other hand, is better explained by aspects of effort and organization in studying, aspects that generally have been found to be interpretable in terms of achievement orientation. Still, in the perception of students, block tests are seen as more rewarding, and progress tests are not taken to be effective means to trigger self-responsible, free and interest based learning, at least not within an assessment system where both progress tests and block tests are summatively used.

Using background knowledge to improve inductive learning of DNA sequences

Abstract: Successful inductive learning requires that training data be expressed in a form where underlying regularities can be recognized by the learning system. Unfortunately, many applications of inductive learning/spl minus/especially in the domain of molecular biology/spl minus/have assumed that data are provided in a form already suitable for learning, whether or not such an assumption is actually justified. This paper describes the use of background knowledge of molecular biology to re-express data into a form more appropriate for learning. Our results show dramatic improvements in classification accuracy for two very different classes of DNA sequences using traditional "off-the-sheIf" decision-tree and neural-network inductive-learning methods. >

Perceptually Grounded Language Learning: Part 2—DETE: A Neural/Procedural Model

Abstract: In Part 1 of this two-part series, we introduced Katamic memory—a neural network architecture capable of robust sequence learning and recognition. In Part 2, we introduce the Blobs World taskjdomain for language learning and describe the DETE language learning system, which is composed of over 50 Katamic memory modules. DETE currently learns small subsets of English and Spanish via association with perceptual! motor inputs. In addition to Kaiamic memory, DETE employs several other novel features: (1) use of feature planes, to encode visual shapes, spatial relationships and the motions of objects, (2) phase-locking of neural firing, in order to represent focus of atention and to bind objects across multiple feature planes, and (3) a method for encoding temporal relationships, so that DETE can learn utterances involving the immediate past and future. We compare DETE to related models and discuss the implications of this approach for language-learning research.

Information processing and the learning context : an analysis from recent perspectives in cognitive psychology

Abstract: Biggs (1993) asked, ‘What framework has more to offer researchers, teachers, and staff developers: that derived from IP, or from SAL?’ His response was to argue that the Student Approaches to Learning perspective best accounts for the context specific motive and strategy components in students' approaches to learning, and their relationship to students' intentions, the teaching/learning context, and the quality of the learning outcome. In contrast, we believe that cross-fertilisation between the Information Processing and SAL perspectives can make a valuable contribution to the understanding of student learning. To illustrate this point, we examine how two more recent IP theories, namely Transfer Appropriate Processing theory and the item and relational information distinction, can inform our understanding of student learning. The major implications of these theories for the conceptualisation of student approaches to learning are: (a) student learning strategies are better classified according to whether they are ‘task-appropriate’ or ‘task-inappropriate’, rather than by the terms ‘deep’ or ‘surface’ strategies; and (b) students will achieve the maximum benefit from a combination of approaches to learning.

Exploring the Statistical Derivation of Transformational Rule Sequences for Part-of-Speech Tagging

Abstract: Eric Brill in his recent thesis (1993b) proposed an approach called "transformation-based error-driven learning" that can statistically derive linguistic models from corpora, and he has applied the approach in various domains including part-of-speech tagging (Brill, 1992; Brill, 1994) and building phrase structure trees (Brill, 1993a). The method learns a sequence of symbolic rules that characterize important contextual factors and use them to predict a most likely value. The search for such factors only requires counting various sets of events that actually occur in a training corpus, and the method is thus able to survey a larger space of possible contextual factors than could be practically captured by a statistical model that required explicit probability estimates for every possible combination of factors. Brill's results on part-of-speech tagging show that the method can outperform the HMM techniques widely used for that task, while also providing more compact and perspicuo.s models. Decision trees are an established learning technique that is also based on surveying a wide space of possible factors and repeatedly selecting a most significant factor or combination of factors. After briefly describing Brill's approach and noting a fast implementation of it, this paper analyzes it in relation to decision trees. The contrast highlights the kinds of applications to which rule sequence learning is especially suited. We point out how it, ma.ages to largely avoid difficulties with overtraining, and show a way of recording the dependencies bt.tween rules in the learned sequence. The analysis throughout is based on part-of-speech tagging experiments using the tagged Brown Corpus (Francis and K.eera, 1979) and a tagged Septuagint Greek version of the first five books of the Bible (CATSS, 1991). Brill 's Approach

Biological learning and artificial intelligence

Abstract: It was once taken for granted that learning in animals and man could be explained with a s imple set of general learning rules, but over the last hundred years, a substantial amount of evidence has been accumulated that points in a quite different direction. In animal learning theory, the laws of learning are no longer considered general. Instead, it has been necessary to explain behaviour in terms of a large set of interacting learning mechanisms and innate behaviours. Artificial intelligence is now on the edge of making the transition from general theories to a view of intelligence that is based on an amalgamate of interacting systems. In the light of the evidence from animal learning theory, such a transition is to be highly desired.

Modelling the Perception of Musical Sequences with Self-organizing Neural Networks

Abstract: A brief review of studies into the psychology of melody perception leads to the conclusion that melodies are represented in long-term memory as sequences of specific items, either intervals or scale notes; the latter representation is preferred. Previous connectionist models of musical-sequence learning are discussed and criticized as models of perception. The Cohen— Grossberg masking field (Cohen & Grossberg, 1987) is described and it is shown how it can be used to generate melodic expectations when incorporated within an adaptive resonance architecture. An improved formulation, the SONNET 1 network (Nigrin, 1990, 1992), is described in detail and modifications are suggested. The network is tested on its ability to learn short melodic phrases taken from a set of simple melodies, before being applied to the learning of the melodies themselves. Mechanisms are suggested for sequence recognition and sequence recall. The advantages of this approach to sequence learning are discussed.

Learning in experimental games

Abstract: Experimental games typically involve subjects playing the same game a number of times. In the absence of perfect rationality by all players, the subjects may use the behavior of their opponents in early rounds to learn about the extent of irrationality in the population they face. This makes the problem of finding the Bayes-Nash equilibrium of the experimental game much more complicated than finding the game-theoretic solution to the ideal game without irrationality. We propose and implement a computationally intensive algorithm for finding the equilibria of complicated games with irrationality via the minimization of an appropriate multi-variate function. We propose two hypotheses about how agents learn when playing experimental games. The first posits that they tend to learn about each opponent as they play it repeatedly, but do not learn about the population parameters through their observations of random opponents (myopic learning). The second posits that both types of learning take place (sequential learning). We introduce a computationally intensive sequential procedure to decide on the informational value of conducting additional experiments. With the help of that procedure, we decided after 12 experiments that our original model of irrationality was unsatisfactory for the purpose of discriminating between our two hypotheses. We changed our models, allowing for two different types of irrationality, reanalyzed the old data, and conducted 7 more experiments. The new model successfully discriminated between our two hypotheses about learning. After only 7 more experiments, our approximately optimal stopping rule led us to stop sampling and accept the model where both types of learning occur

Sequential and parallel neural network vector quantizers

Abstract: Presents novel sequential and parallel learning techniques for codebook design in vector quantizers using neural network approaches. These techniques are used in the training phase of the vector quantizer design. These learning techniques combine the split-and-cluster methodology of the traditional vector quantizer design with neural learning, and lead to better quantizer design (with fewer distortions). The sequential learning approach overcomes the code word underutilization problem of the competitive learning network. As a result, this network only requires partial or zero updating, as opposed to full neighbor updating as needed in the self organizing feature map. The parallel learning network, while satisfying the above characteristics, also leads to parallel learning of the codewords. The parallel learning technique can be used for faster codebook design in a multiprocessor environment. It is shown that this sequential learning scheme can sometimes outperform the traditional LBG algorithm, while the parallel learning scheme performs very close to the LGB and the sequential learning algorithms. >

The representation of structure in sequence prediction tasks

Abstract: Is knowledge acquired implicitly abstract or based on memory for exemplars? This question is at the heart of a current, but long-standing, controversy in the field of implicit learning (see Reber, 1989, for a review). For some authors, implicit knowledge is best characterized as rulelike. For others, however, knowledge acquired implicitly is little more than knowledge about memorized exemplars, or at best, knowledge about elementary features of the material, such as the frequency of particular events. In this paper, I argue that the debate may be ill-posed, and that the two positions are not necessarily incompatible. Using simulation studies, I show that abstract knowledge about the stimulus material may emerge through the operation of elementary, associationist learning mechanisms of the kind that operate in connectionist networks. I focus on a sequence learning task first proposed by Kushner, Cleeremans & Reber (1991), during which subjects are exposed to random fixed-length sequences and are asked to predict the location at which the last element of each sequence will appear. Unknown to them, the location of the last element is determined based on the relationship between specific previous elements. This situation is thus quite complex, because the relevant information is relational, and because it is embedded in a large number of irrelevant contexts. Kushner, Cleeremans & Reber (1991) showed that human subjects are able to learn this material despite limited ability to verbalize their knowledge. In this paper, I first present simulation studies in which connectionist networks are trained to predict the last event of the sequences in the same conditions as subjects were. I focus on issues of representation and transfer. What knowledge do the networks acquire about the temporal extent of the material? What is the form of this knowledge? The results highlight limitations of two well-known models of sequential processing, that is, the SRN model (Cleeremans & McClelland, 1991) and Jordan’s recurrent network (Jordan, 1986), and indicate that a simple decay-based, buffer network may be sufficient to account for human performance. Next, I explore how well the model can transfer to various test situations, in which new sequences may include either relevant or irrelevant Representation of Structure 3 sequence elements that have never been presented during training. I discuss the results in light of the abstraction/memory for instances debate. Based on these and other results, I suggest that the kind of representations developed by connectionist models are intermediate between abstract representations and exemplar-based representations, and that these two extreme forms of representation are points on a continuum. Representation of Structure 4

Chapter 9 – Machine Learning

Abstract: Publisher Summary This chapter focuses on machine learning involving in the artificial intelligence. Various forms of machine learning exist, as indicated by the multitude of terms such as inductive inference, concept formation, explanation-based learning, case-based learning, learning by exploration, PAC learning, reinforcement learning, connectionist learning, and genetic algorithms. Machine learning is a multifaceted and very active area. The chapter discusses some of the issues involved in implementing two main learning techniques: (1) inductive inference, both in general and in the special case of decision trees, and (2) explanation-based learning. Both over- and undergeneralization are corrected on the basis of further data. The process of repeatedly modifying and refining previous hypotheses of the concept on the basis of examples and counterexamples is called inductive inference. The notion of a concept hierarchy is very general, and it is further specialized by considering the actual representation employed.

Pattern recognition dictionary generating method

Abstract: PURPOSE:To efficiently generate a pattern recognition dictionary whose performance is satisfactory by using a shielding means for shielding a specific part of the pattern recognition dictionary from other category in its periphery. CONSTITUTION:This method is constituted of a phase 1 for selecting a learning category and a phase 2 for sequential learning. The phase 1 consists of an initial dictionary generating part 01, a pattern recognizing part 02, a learning category set selecting part 03 and a shielding category set selecting part 04. The phase 2 consists of a learning pattern selecting part 05, a pattern recognizing part 06, a dictionary correction deciding part 07, a dictionary correction executing part 08, and an end deciding part 09. Also, in the recognition dictionary, only a necessary part of learning becomes an object of learning, and a necessary part of learning is shielded by a category having possibility of erroneous recognition, or a necessary part of learning is shielded by an artificial pattern, therefore, even if sequential learning is executed in a part of the dictionary, it is suppressed to the minimum that adverse influence is exerted on the whole dictionary.

On the learning dynamics of spatiotemporal neural networks

Abstract: Previously, spatiotemporal neural networks (STNNs) have been tested for applications such as speech recognition, radar and sonar echoes. STNNs have shown their plausibility using Kohonen's competitive learning and the Kosko/Klopf rule. This paper presents a modified version of the dynamic equation (used in determining the next output of a neuron) that can help ease the tuning problem. For asymmetric or temporal sequence learning, the authors analyze the Kosko/Klopf rule, and prove that the necessary condition in achieving asymptotic stability is to keep 0 >

Temporal pattern generation based on anticipation

Abstract: A neural network model of complex temporal pattern generation is proposed and investigated analytically and by computer simulation Temporal pattern generation is based on recognition of the contexts of individual components Based an its acquired experience the model actively yields system anticipation, which then compares with the actual input flow A mismatch triggers self-organization of context learning, which ultimately leads to resolving various ambiguities in producing complex temporal patterns We show analytically that the network model can learn to generate any complex temporal pattern Multiple patterns can be acquired sequentially by the system, manifesting a form of retroactive interference The model is consistent with cognitive studies of sequential learning >

Integrating induction & instruction: connectionist advice taking

Abstract: Humans improve their performance by means of a variety of learning strategies, including both gradual statistical induction from experience and rapid incorporation of advice. In many learning environments, these strategies may interact in complementary ways. The focus of this work is on cognitively plausible models of multistrategy learning involving the integration of inductive generalization and learning \by being told". Such models might be developed by starting with an architecture for which advice taking is relatively easy, such as one based upon a sentential knowledge representation, and subsequently adding some form of inductive learning mechanism. Alternatively, such models might be grounded in a statistical learning framework appropriately extended to operational-ize instruction. This latter approach is taken here. Speciically, connectionist back-propagation networks (Rumelhart, McClelland, & the PDP Research Group 1986) are made to instantaneously modify their behavior in response to quasi-linguistic advice. Many of the previous approaches to the instruction of connectionist networks have involved the encoding of symbolic rules as initial connection weights which may be later reened by inductive learning (Giles & Omlin 1993) (Tresp, Hollatz, & Ahmad 1993). A major drawback of this approach is that advice may only be given before inductive training begins. This is an unreasonable constraint for a cognitive model of instructed learning. Instead, a connectionist network is needed which may have its behavior altered by a stream of encoded instructions without a delay period for lengthy retraining. The approach which is examined here focuses on encoding the receipt of instruction as motion in a net-work's activation space. In short, advice is presented to such an instructable network as a temporal sequence of instruction tokens, where each token is encoded as an input activation pattern. The network is trained to appropriately modulate its behavior based on input of such advice sequences. The correct interpretation and operationalization of input instruction sequences is learned inductively, but, once this initial learning is complete, instruction following proceeds at the speed of activation propagation. This focus on activation space dynamics allows instructional learning and standard connectionist inductive learning to function in tandem. This strategy has been successfully applied to a simple discrete mapping task and to the learning of natural number arithmetic. In this latter domain, the connectionist adder of Cottrell and Tsung (Cottrell & Tsung 1993), which is capable of systematically operating on arbitrarily large natural numbers, was augmented to receive instruction in various methods of addition and subtraction. …