Theory-based Bayesian models of inductive reasoning 2 Theory-based Bayesian models of inductive reasoning

Since the experiments of Saffran et al. (1996a), there has been a great deal of interest in the question of how statistical regularities in the speech stream might be used by infants to begin to identify individual words. In this work, we use computational modeling to explore the effects of different assumptions the learner might make regarding the nature of words – in particular, how these assumptions affect the kinds of words that are segmented from a corpus of transcribed child-directed speech. We develop several models within a Bayesian ideal observer framework, and use them to examine the consequences of assuming either that words are independent units, or units that help to predict other units. We show through empirical and theoretical results that the assumption of independence causes the learner to undersegment the corpus, with many two- and three-word sequences (e.g. what’s that, do you, in the house) misidentified as individual words. In contrast, when the learner assumes that words are predictive, the resulting segmentation is far more accurate. These results indicate that taking context into account is important for a statistical word segmentation strategy to be successful, and raise the possibility that even young infants may be able to exploit more subtle statistical patterns than have usually been considered. 1

Inductive learning is impossible without overhypotheses, or constraints on the hypotheses considered by the learner. Some of these overhypotheses must be innate, but we suggest that hierarchical Bayesian models help explain how the rest can be acquired. To illustrate this claim, we develop models that acquire two kinds of overhypotheses — overhypotheses about feature variability (e.g. the shape bias in word learning) and overhypotheses about the grouping of categories into ontological kinds like objects and substances.

Situated, spontaneous speech may be ambiguous along acoustic, lexical, grammatical and semantic dimensions. To understand such a seemingly difficult signal, we propose to model the ambiguity inherent in acoustic signals and in lexical and grammatical choices using compact, probabilistic representations of multiple hypotheses. To resolve semantic ambiguities we propose a situation model that captures aspects of the physical context of an utterance as well as the speaker’s intentions, in our case represented by recognized plans. In a single, coherent Framework for Understanding Situated Speech (FUSS) we show how these two influences, acting on an ambiguous representation of the speech signal, complement each other to disambiguate form and content of situated speech. This method produces promising results in a game playing environment and leaves room for other types of situation models.

Address correspondence to

Abstract not found

Word learning is a ‘‘chicken and egg’’ problem. If a child could understand speakers ’ utterances, it would be easy to learn the meanings of individual words, and once a child knows what many words mean, it is easy to infer speakers’ intended meanings. To the beginning learner, however, both individual word meanings and speakers ’ intentions are unknown. We describe a computational model of word learning that solves these two inference problems in parallel, rather than relying exclusively on either the inferred meanings of utterances or cross-situational word-meaning associations. We tested our model using annotated corpus data and found that it inferred pairings between words and object concepts with higher precision than comparison models. Moreover, as the result of making probabilistic inferences about speakers’ intentions, our model explains a variety of behavioral phenomena described in the word-learning literature. These phenomena include mutual exclusivity, one-trial learning, cross-situational learning, the role of words in object individuation, and the use of inferred intentions to disambiguate reference.

Our goal is to automatically recognize and enroll new vocabulary in a multimodal interface. To accomplish this our technique aims to leverage the mutually disambiguating aspects of co-referenced, co-temporal handwriting and speech. The co-referenced semantics are spatially and temporally determined by our multimodal interface for schedule chart creation. This paper motivates and describes our technique for recognizing out-of-vocabulary (OOV) terms and enrolling them dynamically in the system. We report results for the detection and segmentation of OOV words within a small multimodal test set. On the same test set we also report utterance, word and pronunciation level error rates both over individual input modes and multimodally. We show that combining information from handwriting and speech yields significantly better results than achievable by either mode alone.

Everyday inductive inferences are often guided by rich background knowledge. Formal models of induction should aim to incorporate this knowledge, and should explain how different kinds of knowledge lead to the distinctive patterns of reasoning found in different inductive contexts. We present a Bayesian framework that attempts to meet both goals and describe four applications of the framework: a taxonomic model, a spatial model, a threshold model, and a causal model. Each model makes probabilistic inferences about the extensions of novel properties, but the priors for the four models are defined over different kinds of structures that capture different relationships between the categories in a domain. Our framework therefore shows how statistical inference can operate over structured background knowledge, and we argue that this interaction between structure and statistics is critical for explaining the power and flexibility of human reasoning.

Language is about symbols, and those symbols must be grounded in the physical world. Children learn to associate language with sensorimotor experiences during their development. In light of this, we first provide a computational account of how words are mapped to their perceptually grounded meanings. Moreover, the main part of this work proposes and implements a computational model of how word learning influences the formation of object categories to which those words refer. This model simulates the bi-directional relationship between word and object category learning: (1) object categorization provides mental representations of meanings that are mapped to words to form lexical items; (2) linguistic labels help object categorization by providing additional teaching signals; and (3) these two learning processes interplay with each other and form a developmental feedback loop. Compared with the method that performs these two tasks separately, our model shows promising improvements in both word-to-world mapping and perceptual categorization, suggesting a unified view of lexical and category learning in an integrative framework. Most importantly, this work provides a cognitively plausible explanation of the mechanistic nature of early word learning and object learning from co-occurring multisensory data.