For an arbitrary finite algebra $\g A = (A, f, 0, 1)$ one defines a double sequence $a(i,j)$ by $a(i,0)\!=\!a(0,j)\! =\! 1$ and $a(i,j) \!= \!f( a(i, j-1) , a(i-1,j) )$.The problem if such recurrent double sequences are ultimately zero is undecidable, even if we restrict it to the class of commutative finite algebras.

Although many studies have provided evidence that abstract knowledge can be acquired in artificial grammar learning, it remains unclear how abstract knowledge can be attained in sequence learning. To address this issue, we proposed a dual simple recurrent network model that includes a surface SRN encoding and predicting the surface properties of stimuli and an abstract SRN encoding and predicting the abstract properties of stimuli. The results of Simulations 1 and 2 showed that the DSRN model can account for (...) learning effects in the serial reaction time task under different conditions, and the manipulation of the contribution weight of each SRN accounted for the contribution of conscious and unconscious processes in inclusion and exclusion tests in previous studies. The results of human performance in Simulation 3 provided further evidence that people can implicitly learn both chunking and abstract knowledge in sequence learning, and the results of Simulation 3 confirmed that the DSRN model can account for how people implicitly acquire the two types of knowledge in sequence learning. These findings extend the learning ability of the SRN model and help understand how different types of knowledge can be acquired implicitly in sequence learning. (shrink)

This paper proposes several recurrent neural network-based models for recognizing requisite and effectuation parts in Legal Texts. Firstly, we propose a modification of BiLSTM-CRF model that allows the use of external features to improve the performance of deep learning models in case large annotated corpora are not available. However, this model can only recognize RE parts which are not overlapped. Secondly, we propose two approaches for recognizing overlapping RE parts including the cascading approach which uses the sequence of BiLSTM-CRF (...) models and the unified model approach with the multilayer BiLSTM-CRF model and the multilayer BiLSTM-MLP-CRF model. Experimental results on two Japan law RRE datasets demonstrated advantages of our proposed models. For the Japanese National Pension Law dataset, our approaches obtained an \ score of 93.27% and exhibited a significant improvement compared to previous approaches. For the Japan Civil Code RRE dataset which is written in English, our approaches produced an \ score of 78.24% in recognizing RE parts that exhibited a significant improvement over strong baselines. In addition, using external features and in-domain pre-trained word embeddings also improved the performance of RRE systems. (shrink)

Recurrent neural network has been widely applied to many sequential tagging tasks such as natural language process and time series analysis, and it has been proved that RNN works well in those areas. In this paper, we propose using RNN with long short-term memory units for server load and performance prediction. Classical methods for performance prediction focus on building relation between performance and time domain, which makes a lot of unrealistic hypotheses. Our model is built based on events, which (...) is the root cause of server performance. We predict the performance of the servers using RNN-LSTM by analyzing the log of servers in data center which contains user’s access sequence. Previous work for workload prediction could not generate detailed simulated workload, which is useful in testing the working condition of servers. Our method provides a new way to reproduce user request sequence to solve this problem by using RNN-LSTM. Experiment result shows that our models get a good performance in generating load and predicting performance on the data set which has been logged in online service. We did experiments with nginx web server and mysql database server, and our methods can been easily applied to other servers in data center. (shrink)

There is increasing interest in how the pupil dynamics of the eye reflect underlying cognitive processes and brain states. Problematic, however, is that pupil changes can be due to non-cognitive factors, for example luminance changes in the environment, accommodation and movement. In this paper we consider how by modeling the response of the pupil in real-world environments we can capture the non-cognitive related changes and remove these to extract a residual signal which is a better index of cognition and performance. (...) Specifically, we utilize sequence measures such as fixation position, duration, saccades, and blink-related information as inputs to a deep recurrent neural network (RNN) model for predicting subsequent pupil diameter. We build and evaluate the model for a task where subjects are watching educational videos and subsequently asked questions based on the content. Compared to commonly-used models for this task, the RNN had the lowest errors rates in predicting subsequent pupil dilation given sequence data. Most importantly was how the model output related to subjects' cognitive performance as assessed by a post-viewing test. Consistent with our hypothesis that the model captures non-cognitive pupil dynamics, we found (1) the model's root-mean square error was less for lower performing subjects than for those having better performance on the post-viewing test, (2) the residuals of the RNN (LSTM) model had the highest correlation with subject post-viewing test scores and (3) the residuals had the highest discriminability (assessed via area under the ROC curve, AUC) for classifying high and low test performers, compared to the true pupil size or the RNN model predictions. This suggests that deep learning sequence models may be good for separating components of pupil responses that are linked to luminance and accommodation from those that are linked to cognition and arousal. (shrink)

Over the past decade, sequence learning has gradually become a central paradigm through which to study implicit learning. In this chapter, we start by briefly summarizing the results obtained with different variants of the sequence learning paradigm. We distinguish three subparadigms in terms of whether the stimulus material is generated either by following a fixed and repeating sequence (e.g., Nissen & Bullemer, 1987), by relying on a complex set of rules from which one can produce several alternative deterministic sequences (...) (e.g., Lewicki, Hill & Bizot, 1988; Stadler, 1989), or by following the output of a probabilistic set of rules such as instantiated by noisy finite-state grammars (Cleeremans & McClelland, 1991; Jiménez, Mendéz & Cleeremans, 1996). Next, we focus on the processes involved in sequence representation and acquisition. We suggest that the sensitivity to the sequential structure observed in the probabilistic subparadigm can only be a result of the acquisition of a representation of the statistical constraints of the material, and that this sensitivity emerges through the operation of mechanisms that are well instantiated by connectionist models such as the Simple Recurrent Network (Elman, 1990; Cleeremans, 1993b). We present new simulation work meant to explore to what extent the model can also account for specific data obtained in a paradigmatic instance of deterministic, rule-based sequence learning task: Lewicki et al. (1988)'s situation. Finally, we report on the results of an experiment that compares learning on otherwise similar deterministic and probabilistic structures, and we show that learning of both types of structures is equivalent only under conditions that maximally hinder explicit acquisition. Taken together, these simulation and experimental data lend support to the claim that implicit learning in all three sequence learning subparadigms can amount to a form of statistical sequence learning. They also suggest that distinguishing among several theories of sequence representation and acquisition may require us to analize the data in great detail. Hopefully, however, some truth can be found in such details.. (shrink)

Are similar, or even identical, mechanisms used in the computational modeling of speech segmentation, serial image processing, and music processing? We address this question by exploring how TRACX2, a recognition‐based, recursive connectionist autoencoder model of chunking and sequence segmentation, which has successfully simulated speech and serial‐image processing, might be applied to elementary melody perception. The model, a three‐layer autoencoder that recognizes “chunks” of short sequences of intervals that have been frequently encountered on input, is trained on the tone intervals (...) of melodically simple French children's songs. It dynamically incorporates the internal representations of these chunks into new input. Its internal representations cluster in a manner that is consistent with “human‐recognizable” melodic categories. TRACX2 is sensitive to both contour and proximity information in the musical chunks that it encounters in its input. It shows the “end‐of‐word” superiority effect demonstrated by Saffran et al. (1999) for short musical phrases. The overall findings suggest that the recursive autoassociative chunking mechanism, as implemented in TRACX2, may be a general segmentation and chunking mechanism, underlying not only word‐ and image‐chunking, but also elementary melody processing. (shrink)

As linguistic competence so clearly illustrates, processing sequences of events is a fundamental aspect of human cognition. For this reason perhaps, sequence learning behavior currently attracts considerable attention in both cognitive psychology and computational theory. In typical sequence learning situations, participants are asked to react to each element of sequentially structured visual sequences of events. An important issue in this context is to determine whether essentially associative processes are sufficient to understand human performance, or whether more powerful learning (...) mechanisms are necessary. To address this issue, we explore how well human participants and connectionist models are capable of learning sequential material that involves complex, disjoint, longdistance contingencies. We show that the popular Simple Recurrent Network model (Elman, 1990), which has otherwise been shown to account for a variety of empirical findings (Cleeremans, 1993), fails to account for human performance in several experimental situations meant to test the model’s specific predictions. In previous research (Cleeremans, 1993) briefly described in this paper, the structure of center-embedded sequential structures was manipulated to be strictly identical or probabilistically different as a function of the elements surrounding the embedding. While the SRN could only learn in the second case, human subjects were found to be insensitive to the manipulation. In the new experiment described in this paper, we tested the idea that performance benefits from “starting small effects” (Elman, 1993) by contrasting two conditions in which the training regimen was either incremental or not. Again, while the SRN is only capable of learning in the first case, human subjects were able to learn in both. We suggest an alternative model based on Maskara & Noetzel’s (1991) Auto-Associative Recurrent Network as a way to overcome the SRN model’s failure to account for the empirical findings.. (shrink)

Day-to-day lives are affected globally by the epidemic coronavirus 2019. With an increasing number of positive cases, India has now become a highly affected country. Chronic diseases affect individuals with no time identification and impose a huge disease burden on society. In this article, an Efficient Recurrent Neural Network with Ensemble Classifier is built using VGG-16 and Alexnet with weighted model to predict disease and its level. The dataset is partitioned randomly into small subsets by utilizing mean-based splitting method. (...) Various models of classifier create a homogeneous ensemble by utilizing an accuracy-based weighted aging classifier ensemble, which is a weighted model’s modification. Two state of art methods such as Graph Sequence Recurrent Neural Network and Hybrid Rough-Block-Based Neural Network are used for comparison with respect to some parameters such as accuracy, precision, recall, f1-score, and relative absolute error. As a result, it is found that the proposed ERNN-EC method accomplishes accuracy of 95.2%, precision of 91%, recall of 85%, F1-score of 83.4%, and RAE of 41.6%. (shrink)

Sibley et al. (2008) report a recurrent neural network model designed to learn wordform representations suitable for written and spoken word identification. The authors claim that their sequence encoder network overcomes a key limitation associated with models that code letters by position (e.g., CAT might be coded as C‐in‐position‐1, A‐in‐position‐2, T‐in‐position‐3). The problem with coding letters by position (slot‐coding) is that it is difficult to generalize knowledge across positions; for example, the overlap between CAT and TOMCAT is lost. Although (...) we agree this is a critical problem with many slot‐coding schemes, we question whether the sequence encoder model addresses this limitation, and we highlight another deficiency of the model. We conclude that alternative theories are more promising. (shrink)

We explore a network architecture introduced by Elman (1988) for predicting successive elements of a sequence. The network uses the pattern of activation over a set of hidden units from time-step 25-1, together with element t, to predict element t + 1. When the network is trained with strings from a particular finite-state grammar, it can learn to be a perfect finite-state recognizer for the grammar. When the network has a minimal number of hidden units, patterns on the hidden units (...) come to correspond to the nodes of the grammar, although this correspondence is not necessary for the network to act as a perfect finite-state recognizer. We explore the conditions under which the network can carry information about distant sequential contingencies across intervening elements. Such information is maintained with relative ease if it is.. (shrink)

The study addresses the cyclically temporal aspect of sequence recognition, storage and recall using the Recurrent Oscillatory Self-Organizing Map (ROSOM), first introduced by Kaipainen, Papadopoulos and Karhu (1997). The unique solution of the network is that oscillatory States are assigned to network units, corresponding to their `readiness-to-fire''. The ROSOM is a categorizer, a temporal sequence storage system and a periodicity detector designed for use in an ambiguous cyclically repetitive environment. As its external input, the model accepts a multidimensional stream (...) of environment-describing feature configurations with implicit periodicities. The output of the model is one or a few closed cycles abstracted from such a stream, mapped as trajectories on a two-dimensional sheet with an organization reminiscent of multi-dimensional scaling. The model''s capabilities are explored with a variety of workbench data. (shrink)

In this contribution, we use the connection between stable polynomials and orthogonal polynomials on the real line to construct sequences of Hurwitz polynomials that are robustly stable in terms of several uncertain parameters. These sequences are constructed by using properties of orthogonal polynomials, such as the well-known three-term recurrence relation, as well as by considering linear combinations of two orthogonal polynomials with consecutive degree. Some examples are presented.

In everyday interaction, people recurrently animate, enact, or report on talk or thought. In this article, enactments of hypothetical, non-narrative talk in advice-relevant sequences are examined, with a focus on their role in modeling desirable stance or conduct. Data consist of interactions in institutional settings, such as performance appraisal interviews, university teaching, and talk show counseling. It is demonstrated how enactments of possible talk are used as devices for hands-on demonstrations of proper or improper conduct in sequences involving (...) orientations to some kind of problematic behavior or stance, which in turn works to make assessments about different types of conduct. The accomplishment of contrasts between desired and undesired conduct is central, and contributes to the assessment of particular behaviors or stances. Different delivery formats of enactments are examined and compared to the action accomplished. It is argued that modeling talk enactments constitute resources for doing implicit criticism and ‘positive socialization’ in interaction, and that through enactments, participants may perform both explicit and implicit moral work. (shrink)

Comparing the relative sensitivity of direct and indirect measures of learning is proposed as the best way to provide evidence for unconscious learning when both conceptual and operative definitions of awareness are lacking. This approach was first proposed by Reingold & Merikle (1988) in the context of subliminal perception. In this paper, we apply it to a choice reaction time task in which the material is generated based on a probabilistic finite-state grammar (Cleeremans, 1993). We show (1) that participants progressively (...) learn about the statistical structure of the stimulus material over training with the choice reaction time task, and (2) that they can use some of this knowledge to predict the location of the next stimulus in a subsequent “generation” task. However, detailed partial correlational analyses of the correspondence between performance during the reaction time task and the statistical structure of the training material showed that large effects remained even when controlling for explicit knowledge as assessed by the generation task. Hence we conclude (1) that at least some of the knowledge expressed through reaction time performance can not be characterized as conscious, and (2) that even when associations are found at a global level of analysis, dissociations can still be obtained when more detailed analyses are conducted. Finally, we also show that participants are limited in the depth of the contingencies they can learn about, and that these limitations are shared by the Simple Recurrent Network model of Cleeremans & McClelland (1991). (shrink)

The detection and categorization of animate motions is a crucial task underlying social interaction and perceptual decision making. Neural representations of perceived animate objects are partially located in the primate cortical region STS, which is a region that receives convergent input from intermediate-level form and motion representations. Populations of STS cells exist which are selectively responsive to specific animated motion sequences, such as walkers. It is still unclear how and to what extent form and motion information contribute to the (...) generation of such representations and what kind of mechanisms are involved in the learning processes. The article develops a cortical model architecture for the unsupervised learning of animated motion sequence representations. We demonstrate how the model automatically selects significant motion patterns as well as meaningful static form prototypes characterized by a high degree of articulation. Such key poses are selectively reinforced during learning through a cross talk between the motion and form processing streams. Furthermore, we show how sequence-selective representations are learned in STS by fusing static form and motion input from the segregated bottom-up driving input streams. Cells in STS, in turn, feed their activities recurrently to their input sites along top-down signal pathways. We show how such learned feedback connections enable predictions about future input as anticipation generated by sequence-selective STS cells. Network simulations demonstrate the computational capacity of the proposed model by reproducing several experimental findings from neurosciences and by accounting for recent behavioral data. (shrink)

Comparing the relative sensitivity of direct and indirect measures of learning is proposed as the best way to provide evidence for unconscious learning when both conceptual and operative definitions of awareness are lacking. This approach was first proposed by Reingold & Merikle (1988) in the context of subliminal perception. In this paper, we apply it to a choice reaction time task in which the material is generated based on a probabilistic finite-state grammar (Cleeremans, 1993). We show (1) that participants progressively (...) learn about the statistical structure of the stimulus material over training with the choice reaction time task, and (2) that they can use some of this knowledge to predict the location of the next stimulus in a subsequent “generation” task. However, detailed partial correlational analyses of the correspondence between performance during the reaction time task and the statistical structure of the training material showed that large effects remained even when controlling for explicit knowledge as assessed by the generation task. Hence we conclude (1) that at least some of the knowledge expressed through reaction time performance can not be characterized as conscious, and (2) that even when associations are found at a global level of analysis, dissociations can still be obtained when more detailed analyses are conducted. Finally, we also show that participants are limited in the depth of the contingencies they can learn about, and that these limitations are shared by the Simple Recurrent Network model of Cleeremans & McClelland (1991). (shrink)

The demonstration of a sequential congruency effect in sequence learning has been offered as evidence for control processes that act to inhibit automatic response tendencies via unconscious conflict monitoring. Here we propose an alternative interpretation of this effect based on the associative learning of chains of sequenced contingencies. This account is supported by simulations with a Simple Recurrent Network, an associative model of sequence learning. We argue that the control- and associative-based accounts differ in their predictions concerning the magnitude (...) of the sequential congruency effect across training. These predictions are tested by reanalysing data from a study by Shanks, Wilkinson, and Channon . The results support the associative learning account which explains the sequential congruency effect without appealing to control processes. (shrink)

Meetings are complex institutional events at which participants recurrently negotiate institutional roles, which are oriented to, renegotiated, and sometimes challenged. With a view to gaining further understanding of the ongoing negotiation of roles at meetings, this article examines one specific recurring feature of meetings: the act of proposing future action. Based on microanalysis of video recordings of two-party strategy meetings, the study shows that participants orient to at least two aspects when making proposals: 1) the acceptance or rejection of the (...) proposal; and 2) questions of entitlement: who is entitled to launch a proposal, and who is entitled to accept or reject it? The study argues that there is a close interrelation between questions of entitlement, aligning and affiliating moves, and the negotiation of institutional roles. The multimodal analysis also reveals the use of various embodied practices by participants for the local negotiation of entitlement and institutional roles. (shrink)

There is limited knowledge about cancer patients' experiences of uncertainty while waiting for genome sequencing results, and whether prolonged uncertainty contributes to psychological factors in this context. To investigate uncertainty in patients with a cancer of likely hereditary origin while waiting for genome sequencing results, we collected questionnaire and interview data at baseline, and at three and 12 months follow up. Participants had negative attitudes towards uncertainty at baseline, and low levels of uncertainty at three and 12 months. Uncertainty about (...) genome sequencing did not change significantly over time [t = 0.660, p = 0.510]. Greater perceived susceptibility for cancer [r = 0.14, p < 0.01], fear of cancer recurrence [r = 0.19, p < 0.01], perceived importance of genome sequencing [r = 0.24, p < 0.01], intention to change behavior if a gene variant indicating risk is found [r = 0.29, p < 0.01], perceived ability to cope with results [r = 0.36, p < 0.01], and satisfaction with decision to have genome sequencing [r = 0.52, p < 0.01] were significantly correlated with negative attitudes towards uncertainty at baseline. Multiple primary cancer diagnoses [B = −2.364 [−4.238, −0.491], p = 0.014], lower perceived ability to cope with results [B = −0.1.881 [−3.403, −0.359], p = 0.016] at baseline, greater anxiety about genome sequencing [B = 0.347 [0.148, 0.546], p = 0.0012] at 3 months, and greater perceived uncertainty about genome sequencing [B = 0.494 [0.267, 0.721] p = 0.000] at 3 months significantly predicted greater perceived uncertainty about genome sequencing at 12 months. Greater perceived uncertainty about genome sequencing at 3 months significantly predicted greater anxiety about genome sequencing at 12 months [B = 0.291 [0.072, 0.509], p = 0.009]. Semi-structured interviews revealed that while participants were motivated to pursue genome sequencing as a strategy to reduce their illness and risk uncertainty, genome sequencing generated additional practical, scientific and personal uncertainties. Some uncertainties were consistently discussed over the 12 months, while others emerged over time. Similarly, some uncertainty coping strategies were consistent over time, while others emerged while patients waited for their genome sequencing results. This study demonstrates the complexity of uncertainty generated by genome sequencing for cancer patients and provides further support for the inter-relationship between uncertainty and anxiety. Helping patients manage their uncertainty may ameliorate psychological morbidity. (shrink)

American Academy of Pediatrics (AAP) and American College of Medical Genetics (ACMG) recently provided two recommendations about predictive genetic testing of children. The Clinical Sequencing Exploratory Research Consortium's Pediatrics Working Group compared these recommendations, focusing on operational and ethical issues specific to decision making for children. Content analysis of the statements addresses two issues: (1) how these recommendations characterize and analyze locus of decision making, as well as the risks and benefits of testing, and (2) whether the guidelines conflict or (...) come to different but compatible conclusions because they consider different testing scenarios. These statements differ in ethically significant ways. AAP/ACMG analyzes risks and benefits using best interests of the child and recommends that, absent ameliorative interventions available during childhood, clinicians should generally decline to order testing. Parents authorize focused tests. ACMG analyzes risks and benefits using the interests of the child and other family members and recommends that sequencing results be examined for additional variants that can lead to ameliorative interventions, regardless of age, which laboratories should report to clinicians who should contextualize the results. Parents must accept additional analysis. The ethical arguments in these statements appear to be in tension with each other. (shrink)

Learning the structure of event sequences is a ubiquitous problem in cognition and particularly in language. One possible solution is to learn a probabilistic generative model of sequences that allows making predictions about upcoming events. Though appealing from a neurobiological standpoint, this approach is typically not pursued in connectionist modeling. Here, we investigated a sequential version of the restricted Boltzmann machine, a stochastic recurrent neural network that extracts high-order structure from sensory data through unsupervised generative learning and (...) can encode contextual information in the form of internal, distributed representations. We assessed whether this type of network can extract the orthographic structure of English monosyllables by learning a generative model of the letter sequences forming a word training corpus. We show that the network learned an accurate probabilistic model of English graphotactics, which can be used to make predictions about the letter following a given context as well as to autonomously generate high-quality pseudowords. The model was compared to an extended version of simple recurrent networks, augmented with a stochastic process that allows autonomous generation of sequences, and to non-connectionist probabilistic models. We conclude that sequential RBMs and stochastic simple recurrent networks are promising candidates for modeling cognition in the temporal domain. (shrink)

An account of the development of temporal understanding is proposed which links such understanding with the development of episodic memory. We distinguish between different ways of representing time in terms of the kinds of temporal frameworks they involve. Distinctions are made between frameworks that are perspectival or nonperspectival and those that represent recurrent sequences or particular times. Even primitive temporal understanding integrates both perspectival and nonperspectival components. However, since early frameworks are event-based and localized, they are not yet (...) sufficient for episodic memory in that they do not enable the child to think of events as having occurred at particular points in time. We describe the emergence of new kinds of frameworks in terms of the development of temporal decentering. Two levels of temporal decentering are distinguished, with the higher level involving an appreciation of how event representations depend on one's temporal perspective. (shrink)

Call a (strictly increasing) sequence (rn) of natural numbers regular if it satisfies the following condition: rn+1/rn→θ∈R>1∪{∞} and, if θ is algebraic, then (rn) satisfies a linear recurrence relation whose characteristic polynomial is the minimal polynomial of θ. Our main result states that (Z,+,0,R) is superstable whenever R is enumerated by a regular sequence. We give two proofs of this result. One relies on a result of E. Casanovas and M. Ziegler and the other on a quantifier elimination result. We (...) also show that (Z,+,0,<,R) is NIP whenever R is enumerated by a regular sequence that is ultimately periodic modulo m for all m>1. (shrink)

Expressions are synonymous if they have the same semantic content. Complex expressions are synonymously isomorphic in Alonzo Church’s sense if one is obtainable from the other by a sequence of alphabetic changes of bound variables or replacements of component expressions by syntactically simple synonyms. Synonymous isomorphism provides a very strict criterion for synonymy of sentences. Several eminent philosophers of language hold that synonymous isomorphism is not strict enough. These philosophers hold that ‘Greeks prefer Greeks’ and ‘Greeks prefer Hellenes’ express different (...) propositions even if they are synonymously isomorphic. They hold that the very recurrence (multiple occurrence) of ‘Greeks’ contributes to the proposition expressed something that indicates the very recurrence in question. Kit Fine argues that this thesis, which he labels semantic relationism calls for a radically new conception of semantics. I have argued that the relevant phenomenon is wholly pragmatic, entirely non-semantic. Here I supplement the case with a new argument. No cognition without recognition—or almost none. With this observation, standard Millianism has sufficient resources to confront Frege’s puzzle and related problems without injecting pragmatic phenomena where they do not belong. (shrink)

We examine two connectionist networks—a fractal learning neural network (FLNN) and a Simple Recurrent Network (SRN)—that are trained to process center-embedded symbol sequences. Previous work provides evidence that connectionist networks trained on infinite-state languages tend to form fractal encodings. Most such work focuses on simple counting recursion cases (e.g., anbn), which are not comparable to the complex recursive patterns seen in natural language syntax. Here, we consider exponential state growth cases (including mirror recursion), describe a new training scheme (...) that seems to facilitate learning, and note that the connectionist learning of these cases has a continuous metamorphosis property that looks very different from what is achievable with symbolic encodings. We identify a property—ragged progressive generalization—which helps make this difference clearer. We suggest two conclusions. First, the fractal analysis of these more complex learning cases reveals the possibility of comparing connectionist networks and symbolic models of grammatical structure in a principled way—this helps remove the black box character of connectionist networks and indicates how the theory they support is different from symbolic approaches. Second, the findings indicate the value of future, linked mathematical and empirical work on these models—something that is more possible now than it was 10 years ago. (shrink)

A recurrent theme that is addressed in psychotherapies is the client’s conflicting emotions. This article discusses discursive practices of working on conflicting emotions during psychodynamic psychotherapy. We focus on a phenomenon that we refer to as a ‘delayed response’ and analyze the client’s uses of interactional means, such as a display of negative experience, to invite affiliation or empathy from the therapist. The therapist, however, does not take a turn in the first possible place after the client’s turn. Recurrently, (...) the therapist’s silence is followed by the client’s new turn that backs down from the emotional experience under discussion. After these retractions, the therapists respond with a turn that is responsive both to the retraction and to the initial display of negative experience that occurred prior to it. We argue that the timing of the therapist’s response in these sequences is in the service of psychotherapeutic work on conflicting emotions. (shrink)

The forms of words as they appear in text and speech are central to theories and models of lexical processing. Nonetheless, current methods for simulating their learning and representation fail to approach the scale and heterogeneity of real wordform lexicons. A connectionist architecture termed thesequence encoderis used to learn nearly 75,000 wordform representations through exposure to strings of stress‐marked phonemes or letters. First, the mechanisms and efficacy of the sequence encoder are demonstrated and shown to overcome problems with traditional slot‐based (...) codes. Then, two large‐scale simulations are reported that learned to represent lexicons of either phonological or orthographic wordforms. In doing so, the models learned the statistics of their lexicons as shown by better processing of well‐formed pseudowords as opposed to ill‐formed (scrambled) pseudowords, and by accounting for variance in well‐formedness ratings. It is discussed how the sequence encoder may be integrated into broader models of lexical processing. (shrink)

The forms of words as they appear in text and speech are central to theories and models of lexical processing. Nonetheless, current methods for simulating their learning and representation fail to approach the scale and heterogeneity of real wordform lexicons. A connectionist architecture termed thesequence encoderis used to learn nearly 75,000 wordform representations through exposure to strings of stress‐marked phonemes or letters. First, the mechanisms and efficacy of the sequence encoder are demonstrated and shown to overcome problems with traditional slot‐based (...) codes. Then, two large‐scale simulations are reported that learned to represent lexicons of either phonological or orthographic wordforms. In doing so, the models learned the statistics of their lexicons as shown by better processing of well‐formed pseudowords as opposed to ill‐formed (scrambled) pseudowords, and by accounting for variance in well‐formedness ratings. It is discussed how the sequence encoder may be integrated into broader models of lexical processing. (shrink)

Observational learning can enhance the acquisition and performance quality of complex motor skills. While an extensive body of research has focused on the benefits of synchronous (i.e., concurrent physical practice) and non-synchronous (i.e., delayed physical practice) observational learning strategies, the question remains as to whether these approaches differentially influence performance outcomes. Accordingly, we investigate the differential outcomes of synchronous and non-synchronous observational training contexts using a novel dance sequence. Using multidimensional cross-recurrence quantification analysis, movement time-series were recorded for novice dancers (...) who either synchronised with (n = 22) or observed and then imitated (n = 20) an expert dancer. Participants performed a 16-count choreographed dance sequence for 20 trials assisted by the expert, followed by one final, unassisted performance trial. Although end-state performance did not significantly differ between synchronous and non-synchronous learners, a significant decline in performance quality from imitation to independent replication was shown for synchronous learners. A non-significant positive trend in performance accuracy was shown for non-synchronous learners. For all participants, better imitative performance across training trials led to better end-state performance, but only for the accuracy (and not timing) of movement reproduction. Collectively, the results suggest that synchronous learners came to rely on a realtime mapping process between visual input from the expert and their own visual and proprioceptive intrinsic feedback, to the detriment of learning. Thus, the act of synchronising alone does not ensure an appropriate training context for advanced sequence learning. (shrink)

We present results from numerical studies of supervised learning operations in recurrent networks considered as graphs, leading from a given set of input conditions to predetermined outputs. Graphs that have optimized their output for particular inputs with respect to predetermined outputs are asymptotically stable and can be characterized by attractors which form a representation space for an associative multiplicative structure of input operations. As the mapping from a series of inputs onto a series of such attractors generally depends on (...) the sequence of inputs, this structure is generally noncommutative. Moreover, the size of the set of attractors, indicating the complexity of learning, is found to behave non-monotonically as learning proceeds. A tentative relation between this complexity and the notion of pragmatic information is indicated. (shrink)

The accurate recognition of speech is beneficial to the fields of machine translation and intelligent human–computer interaction. After briefly introducing speech recognition algorithms, this study proposed to recognize speech with a recurrent neural network (RNN) and adopted the connectionist temporal classification (CTC) algorithm to align input speech sequences and output text sequences forcibly. Simulation experiments compared the RNN-CTC algorithm with the Gaussian mixture model–hidden Markov model and convolutional neural network-CTC algorithms. The results demonstrated that the more training (...) samples the speech recognition algorithm had, the higher the recognition accuracy of the trained algorithm was, but the training time consumption increased gradually; the more samples a trained speech recognition algorithm had to test, the lower the recognition accuracy and the longer the testing time. The proposed RNN-CTC speech recognition algorithm always had the highest accuracy and the lowest training and testing time among the three algorithms when the number of training and testing samples was the same. (shrink)

In lieu of an abstract, here is a brief excerpt of the content:THIS MESSAGE IS FOR YOU. MAYBE. by Joseph Agassi There is a mood often enough conjured in science fiction literature to be familiar to every fan, the mood of seemingly intentional yet probably remdom contact between two individuals across immense space-time expanses. The hero of a complicated chase story has lost contact with the mother planet, has long ago leuided on a strange pleuiet, emd there, right now, just (...) walking across the plaza he sees a huge poster, with a piece of commercial advertisement or some similar message of purely local import printed on it. The poster is decorated with an arabesque which puzzles him, quickens his pulse, intrigues him, fascinates him — until it hits him insteinteuneously in full force: the arabesque is a message, written in his own native language, telling him how to reestablish contact. The story, as I say, has many veurietnts. And it has, it seems, a very widespread appeal. It is couched in terms of adventure stories, but its mood is not diat of adventure. The mood is more of die one associated with the idea of the eterned return, die idea of timelessness or of time standing still (and space vanishing), the idea the western philosophies usually associate with the ideas of Nicholas, the Renaissance mystic bishop ofCusa, though it is a recurrent idea in many cultures, of the identity of the vanishingly smedl and the infinitely large. Indeed, the meeting of the message and the recipient is symbolized by the meeting of the infinite line and the point of vanishing length. Whence this universality? I suppose it is a fact of biological nature. Think of the scent a living thing, be it a fish or em emtelope, leaves behind to contact a possible mate. It is a few complex molecules in a sea of molecules of astronomical numbers: just this room contains more molecules than you can imagine. And the animal follows a long tortuous path to find its mate. According to Willietm Craig and Konrad Lorenz, an animal does not have the slightest clue what it is after. Even while building a nest it has no complete 95 96Philosophy and Literature picture of its task. In each of its stages it goes after diverse kinds of things, yet it knows not which things, let alone their inborn rigidly fixed sequence. Yet the animal in charge of a tetsk, whether of finding a mate or of finding a twig for a nest, will be resdess — say Craig and Lorenz — and move about, and try to read the message in any piece of arabesque, until one day one piece of arabesque will med«: sense and enable its finder to complete die next step in the mysterious complex task and open up die search for the next message, the next step further along die unfolding series of preordained ordeals that life is strewn widi. Consider writing proper, like my writing this message for you right now. We do not know the background to literacy, much less its actual origins, except that these were far from unique: though many cultures eire illiterate or have acquired literacy from more evolved neighbors, mere eire many places on earth where literacy evolved independently. Yet even here we see a sequence: many independent ideographic scripts, some ofthem evolving into syllabic scripts and one or two of these evolving into an alphabet proper. How and why, we do not know. There is room for speculation, though. It is hard to imagine how the very first writer conceived of the very first message he wrote without consulting his recipient first. There are a few possible explemations here. First, that writer and recipient are one: writing was a form ofmemorizing, or ofconsigning matters to memory more stable than that of the human brain — the granite rock, for instance. The ancient Egyptian god of literacy was the god of memory. Primitive shepherds use knots in their rope and cuts on their sticks as aides memoire, as bookkeeping of sorts, and scholars consider these the origins of writings. This is evidenced by the runic writings of the Norse, whose letters are images of sticks with cuts on them. Yet runic writing is alphabetic and... (shrink)

The symbol acquisition and manipulation abilities are one of the inherent characteristics of human beings comparing with other creatures. In this paper, based on recurrent self-organizing map and dynamics-based information processing system, we propose a dynamics based self-organizing map. This method enables designing a topological map using time sequence data, which causes recognition and generation of the robot motion. Using this method, we design the self-organizing symbol acquisition system and robot motion generation system for a humanoid robot. By implementing (...) DBSOM to the robot in the real world, we realize the symbol acquisition from the experimental data and investigate the spatial property of the obtained DBSOM. (shrink)

This paper presents a theoretical framework of the circular shift network coding system through the study of nonmultiple clustered interval music performance style conversion and the analysis of music conversion by using circular shift topology, and a series of basic research results of circular shift network coding is obtained under this framework. It reveals the essential connection between scalar network coding based on finite domain and cyclic shift network coding, designs a solution construction algorithm for cyclic shift network coding under (...) multicast network, and portrays the multicast capacity of cyclic shift network coding. It overcomes the problem that the piano roll-curtain representation cannot distinguish between a single long note and multiple consecutive notes of the same pitch, describes musical information more comprehensively, extracts musical implicit style from the note matrix based on autoencoder, and better eliminates the potential influence of musical content on musical performance style. A two-way recurrent neural network based on the gated recurrent unit is used to extract a sequence of note feature vectors of different styles, and a one-dimensional convolutional neural network is used to predict the intensity of the extracted note feature vector sequence for a specific style, which better learns the intensity variation of different styles of MIDI music. (shrink)

This paper discusses a series of case studies on observations, experiments, and the theoretical interpretation between 1890 and 1960 of a shift of dark Fraunhofer lines in the solar spectrum. I argue for the use of flow charts to analyze interconnections and to identify sequences of research strategies. Also I advocate using a newly-developed tool called "block diagram" representation of experimental systems as an appropriate method to identify recurrent patterns in the interplay of instrumentation, experiment, and theory in (...) research episodes. (shrink)

Short-term or working memory provides temporary storage of information in the brain after an experience and is associated with conscious awareness. Neurons sensitive to the multiple stimulus attributes comprising an experience are distributed within many brain regions. Such distributed cell assemblies, activated by an event, are the most plausible system to represent the WM of that event. Studies with a variety of imaging technologies have implicated widespread brain regions in the mediation of WM for different categories of information. Each kind (...) of WM may thus be expected to involve many brain regions rather than a local, uniquely dedicated set of cells. Neurons in a distributed “cell assembly” may be self-selected by their temporally coherent activations. The process by which this fragmented representation of the recent past is reassembled to accomplish essentially automatic and reliable recognition of a recurrent event constitutes an important problem. One plausible mechanism to achieve the identification of past with previous events would require that the representational system mediating WM must coexist in spatial extent and somehow overlap in temporal activation with cell ensembles registering input from subsequent events. The detection of such a postulated mechanism required an experimental approach which would focus upon spatial patterns of coherent activation while information about different events was stored in WM and retrieved, rather than focusing upon the temporal sequences of activation in localized regions of interest. For this purpose, the familiar delayed matching from sample task was modified. A series of information-free flashes, or “noncontingent probes,” was presented before an initial series of visual information items, the Priming Sample, which were to be held in WM during a Delay Period. A second series of visual information items were then presented, the Matching Sample. The task required detection of any item in the second series which had beenabsentfrom the initial series. Thirty such trials with a particular category of visual information constituted a single task. Several DMS tasks with this standardized design, but with different categories of visual information, were presented within each test session. The information categories included letters of the alphabet, single digit numbers, or faces from a school yearbook. Event-related potentials , were computed from 21 standardized electrode placements, separately for information-free probes and for information items in each interval of the trials within a task. Because each electrode is particularly sensitive to coherent activation of neurons in the immediately underlying brain regions, topographic maps were constructed and interpolated across the surface of the scalp. The momentary fluctuations of the resulting voltage “landscapes” throughout the task were then subjected to quantitative analysis. Distinctive landscapes sometimes persisted for prolonged periods, implying sustained engagement of very large populations of neurons. “Difference landscapes” were constructed by subtraction of topographic maps evoked by noncontingent probes during the Delay Period from maps of probe ERPs before the presentation of the initial information in the Priming Sample. Such probe difference landscapes displayed recurrent high similarity to momentary landscapes elicited during subsequent presentation of the information items in the Matching Sample. It seemed as if the distributed cell assembly continuously engaged by mediation of WM of the diverse attributes of the initial stimuli was being dynamically compared to the ensembles engaged by registration of the subsequent stimuli. Spatial Principal Component Analysis was applied to the sequences of momentary voltage landscapes observed throughout trials of each task. This method sought a small number of spatial patterns with which these large sets of inhomogeneous spatial distributions of voltage could be reconstructed. This is the spatial analog of the reconstruction of local ERPs by temporal principal components, as often described previously. Five Spatial Principal Components were found which accounted for about 90% of the total variance of voltage across the surface of the scalp throughout every task. Theloadings,or distinguishing topographic features, of these SPCs, were highly similar during every cognitive task for every subject. However,factor scores,or relative average contribution to the overall voltage distributions, of the different SPCs varied substantially among subjects between the tasks and momentarily within successive intervals of each task. These five SPCs may reflect coherent activation of huge distributed ensembles of neurons which comprise independent but interacting functional brain subsystems. These subsystems may correspond to basic resources available to individuals for allocation to mediate conscious evaluation of information during cognitive activity, providing a filter to bind together fractionated representations of the past to evaluate the present. (shrink)

Saffran, Newport, and Aslin (1996a) found that human infants are sensitive to statistical regularities corresponding to lexical units when hearing an artificial spoken language. Two sorts of segmentation strategies have been proposed to account for this early word‐segmentation ability: bracketing strategies, in which infants are assumed to insert boundaries into continuous speech, and clustering strategies, in which infants are assumed to group certain speech sequences together into units (Swingley, 2005). In the present study, we test the predictions of two (...) computational models instantiating each of these strategies i.e., Serial Recurrent Networks: Elman, 1990; and Parser: Perruchet & Vinter, 1998 in an experiment where we compare the lexical and sublexical recognition performance of adults after hearing 2 or 10 min of an artificial spoken language. The results are consistent with Parser’s predictions and the clustering approach, showing that performance on words is better than performance on part‐words only after 10 min. This result suggests that word segmentation abilities are not merely due to stronger associations between sublexical units but to the emergence of stronger lexical representations during the development of speech perception processes. (shrink)

Nietzsche's doctrine of the -eternal recurrence of the same- - the conception that the universe of events repeats itself in the same sequence, to infinity - is often taken to be logically incoherent: if an event recurs, it is not identically the same as the event itself, and if taken as self-identical cannot be the recurrence of anything. This book offers a new interpretation of the doctrine so as to rescue it from the charge of incoherence. It shows that the (...) doctrine is an outgrowth of ideas found in Nietzsche's philosophy of nature, among them that space is Riemannian and that time is relative to events, not an independently existing continuum which underlies events.". (shrink)

The dynamics of neural populations capture the laws of the mindThis paper focuses on the level of neural networks. Examining the case of recurrent neural networks, the paper argues that the dynamics of neural populations form a privileged level of explanation in cognitive science. According to Schöner, this level is privileged, because it enables cognitive scientists to discover the laws governing organisms’ cognition and behaviour.

Polemical writings about philosophers, of little use if directed against straw men as is likely if not based on historical understanding, must incorporate cultural history, which, in focussing on a philosophy's relationship to its age, justifies ignoring historical sequence so long as figures are placed in context. Philosophy does progressively clarify what certain recurrent types of problems involve. The historian-philosopher writing a history of problems must know intimately philosopher and period, and reveal assumptions and aspects of problems hidden to (...) the philosopher himself. Such a history does not merely report philosophy's results but alone elucidates its inner development. (shrink)

ABSTRACT This article reappraises some aspects of Alexander Nehamas's Nietzsche: Life as Literature. It recognizes as strengths of the book Nehamas's emphasis on Nietzsche's mode of writing and his idea that unified selfhood is an exceptional state that is achieved rather than given. However, it takes issue with the claim that Nietzsche holds a superessentialist view of the self. That view is not clearly supported by textual evidence, does not follow from Nietzsche's regarding the self as simply a sequence of (...) experiences and actions, and is not required in order to explain the “all-or-nothing” requirement of the eternal recurrence. The article argues further that unified selfhood cannot be accounted for solely on the model of literary character, given what is now generally recognized as Nietzsche's naturalistic psychology of drives. However, the Nietzschean self need not be viewed as immutable and is amenable to the influence of culture and consciousness in ways that may encourage a positive account of becoming a unified self. (shrink)

How people learn chunks or associations between adjacent items in sequences was modelled. Two previously successful models of how people learn artificial grammars were contrasted: the CCN, a network version of the competitive chunker of Servan‐Schreiber and Anderson [J. Exp. Psychol.: Learn. Mem. Cogn. 16 (1990) 592], which produces local and compositionally‐structured chunk representations acquired incrementally; and the simple recurrent network (SRN) of Elman [Cogn. Sci. 14 (1990) 179], which acquires distributed representations through error correction. The models' susceptibility (...) to two types of interference was determined: prediction conflicts, in which a given letter can predict two other letters that appear next with an unequal frequency; and retroactive interference, in which the prediction made by a letter changes in the second half of training. The predictions of the models were determined by exploring parameter space and seeing howdensely different regions of the space of possible experimental outcomes were populated by model outcomes. For both types of interference, human data fell squarely in regions characteristic of CCN performance but not characteristic of SRN performance. (shrink)

After critical appraisal of mathematical and biological characteristics of the model, we discuss how a classical hippocampal neural network expresses functions similar to those of the chaotic model, and then present an alternative stimulus-driven chaotic random recurrent neural network (RRNN) that learns patterns as well as sequences, and controls the navigation of a mobile robot.

The Random Number Generation task has a long history in neuropsychology as an assessment procedure for executive functioning. In recent years, understanding of human behavior has gradually changed from reflecting a static to a dynamic process and this shift in thinking about behavior gives a new angle to interpret test results. However, this shift also asks for different methods to process random number sequences. The RNG task is suited for applying non-linear methods needed to uncover the underlying dynamics of (...) random number generation. In the current article we present RandseqR: an R-package that combines the calculation of classic randomization measures and Recurrence Quantification Analysis. RandseqR is an easy to use, flexible and fast way to process random number sequences and readies the RNG task for current scientific and clinical use. (shrink)

At the outset of this discussion, I undertook to present an argument from design which would follow Swinburne's example in making use of a priori judgments, while avoiding some of the objections which have been posed in response to his treatment of these issues. So we need to ask: how does this approach to the question of design compare with Swinburne's?Swinburne argues that a chaotic world is a priori more likely than an ordered world: this consideration provides one central reason, (...) on his account, for giving an explanation of some sort for the world's regularity. The other central argument he advances for this claim is the argument from analogy (in terms of the coins) which we noted earlier. The approach I have taken offers an alternative route to this same conclusion. In particular, it substitutes the simpler a priori judgments recorded in (i) and (ii) for the rather difficult and contentious claim that chaos is a priori more likely than order. In place of this claim, I have offered the judgment that order, or recurrence, is more likely given the activity of a common source or common kind of source than otherwise: this proposal does not commit us to a view either way on the question of whether order is a priori likely per se. Moreover, in place of Swinburne's analogical argument, I have offered an a priori approach, with the advantages I have noted.Given that recurrence is to be explained, we might ask: why offer an explanation in terms of design? On this point, Swinburne argues, for instance, that no other explanation of temporal regularity is even possible a priori. Again, the a priori principles which I have used, in (iv) and (v), may be less ambitious, but at the same time more persuasive. In support of this same idea, Swinburne also cites various ideas to do with the predictive power of the idea of design. I have tried to bring out the role of this sort of consideration in terms of my principle (v). Principle (iv) has no place in Swinburne's account, in view of his reliance on the principle of simplicity as a measure of prior probability.Lastly, we may ask: if we are to cite a designer, are there reasons for attributing to this agent more powers than are needed for the production of the effect to be explained? On this point, Swinburne cites the principle of simplicity. Again, my approach avoids what has proved to be a relatively controversial judgment about the nature of a priori probabilities, offering in place of the principle of simplicity the less ambitious principle recorded in (iii). At this point, I have moreover inverted the logical sequence of Swinburne's argument: it seems to me that, in the ways I have indicated, it is helpful to consider the extent of the powers of the source of recurrence before addressing the question of design.In these various ways, I hope I have made good my undertaking to present an argument which avoids some of the controversy surrounding the particular measures of a priori probability which figure in Swinburne's argument. Moreover, I hope that this approach provides an indication of how a priori judgments may function in a relatively unproblematic way within an argument from design, in so far as (i)–(v) are all rather modest proposals. In sum, the argument I have presented is distinguished by its explicit use of the a priori judgments recorded in (i)–(v), by its attempt to buttress in this fashion analogical forms of argument, and by the logical role it gives to the idea that the source of regularity possesses more powers than are required for the production of this effect.Lastly, we might ask: how persuasive is this argument? Of course, the cogency of the idea of design depends upon the balance of debate in other areas of the philosophy of religion, especially upon our ability to provide some account of the existence of evil. In this paper, I have been concerned to argue simply that recurrence by kind provides evidence for design: I have not addressed the question of whether other features of the world provide good evidence against the idea of (benevolent) design. However, if we confine our attention to this one phenomenon, there is it seems to me good evidence for the idea of design, (i) and (ii) suggest that recurrence surely calls for some explanation; (iii), together with the existence in nature of statistical irregularities, suggests that whatever provides this explanation could have brought about other effects besides; and design seems the only clear explanation of why this effect should have been brought about, if (as I have argued) analogies drawn from vegetable and animal reproduction fail, and if we cannot explain the effect satisfactorily by reference to the conditions of observation. Moreover, I have argued that there are reasons for supposing that the probability a priori of design is relatively high in relation to the probability a priori of any rival hypothesis of equivalent predictive power. In brief, this is because the design hypothesis (unlike the hypothesis of theism) can cite an agent of relatively indeterminate power in order to account for the phenomenon to be explained. In this regard, it is less ‘precisely defined’ than any rival hypothesis of equivalent predictive power. If all of this is so, then as philosophers from early times have supposed, temporal regularity provides the basis for a powerful argument in favour of design. It remains true, of course, that its import can be judged in full only when we have taken into account the relevance of other phenomena, many of which are apparently less favourable to the idea of design. (shrink)

Part-of-speech tagging for English composition is the basis for automatic correction of English composition. The performance of the part-of-speech tagging system directly affects the performance of the marking and analysis of the correction system. Therefore, this paper proposes an automatic scoring model for English composition based on article part-of-speech tagging. First, use the convolutional neural network to extract the word information from the character level and use this part of the information in the coarse-grained learning layer. Secondly, the word-level vector (...) is introduced, and the residual network is used to establish an information path to integrate the coarse-grained annotation and word vector information. Then, the model relies on the recurrent neural network to extract the overall information of the sequence data to obtain accurate annotation results. Then, the features of the text content are extracted, and the automatic scoring model of English composition is constructed by means of model fusion. Finally, this paper uses the English composition scoring competition data set on the international data mining competition platform Kaggle to verify the effect of the model. (shrink)