Copyright (c) 2009 University of Pennsylvania All rights reserved. http://repository.upenn.edu/ircs_reports Recent documents in IRCS Technical Reports Series en-us Fri, 13 Nov 2009 23:22:47 PST 3600 http://repository.upenn.edu/ircs_reports/202 http://repository.upenn.edu/ircs_reports/202 Thu, 12 Nov 2009 07:56:14 PST I demonstrate the application of hierarchical regression modeling, a state-of-the-art technique for statistical inference, to language research. First, a stable sociolinguistic variable in Philadelphia (Labov, 2001) is reconsidered, with attention paid to the treatment of collinearities among socioeconomic predictors. I then demonstrate the use of hierarchical models to account for the random sampling of subjects and items in an experimental setting, using data from a study of word-learning in the face of tonal variation (Quam and Swingley, forthcoming). The results from these case studies demonstrate that modeling sampling from the population has empirical consequences. Kyle Gorman http://repository.upenn.edu/ircs_reports/201 http://repository.upenn.edu/ircs_reports/201 Mon, 07 May 2007 12:40:06 PDT The incidence of autism spectrum disorders may have increased in the past few decades. If so, it is important to identify possible causes. Here we consider whether exposure to artificial displays such as television in the first year or two of life is a contributing factor for infants already at risk of autism. Correlational studies and anecdotal evidence from clinical practice are consistent with this hypothesis. Are there also theoretical reasons for concern? Benjamin Backus http://repository.upenn.edu/ircs_reports/200 http://repository.upenn.edu/ircs_reports/200 Wed, 20 Sep 2006 08:31:36 PDT Selectional constraints are limitations on the applicability of predicates to arguments. For example, the statement "The number two is blue" may be syntactically well formed, but at some level it is anomalous -- BLUE is not a predicate that can be applied to numbers. According to the influential theory of (Katz and Fodor, 1964), a predicate associates a set of defining features with each argument, expressed within a restricted semantic vocabulary. Despite the persistence of this theory, however, there is widespread agreement about its empirical shortcomings (McCawley, 1968; Fodor, 1977). As an alternative, some critics of the Katz-Fodor theory (e.g. (Johnson-Laird, 1983)) have abandoned the treatment of selectional constraints as semantic, instead treating them as indistinguishable from inferences made on the basis of factual knowledge. This provides a better match for the empirical phenomena, but it opens up a different problem: if selectional constraints are the same as inferences in general, then accounting for them will require a much more complete understanding of knowledge representation and inference than we have at present. The problem, then, is this: how can a theory of selectional constraints be elaborated without first having either an empirically adequate theory of defining features or a comprehensive theory of inference? In this dissertation, I suggest that an answer to this question lies in the representation of conceptual knowledge. Following Miller (1990b), I adopt a "differential" approach to conceptual representation, in which a conceptual taxonomy is defined in terms of inferential relationships rather than definitional features. Crucially, however, the inferences underlying the stored knowledge are not made explicit. My hypothesis is that a theory of selectional constraints need make reference only to knowledge stored in such a taxonomy, without ever referring overtly to inferential processes. I propose such a theory, formalizing selectional relationships in probabilistic terms: the selectional behavior of a predicate is modeled as its distributional effect on the conceptual classes of its arguments. This is expressed using the information-theoretic measure of relative entropy (Kullback and Leibler, 1951), which leads to an illuminating interpretation of what selectional constraints are: the strength of a predicate's selection for an argument is identified with the quantity of information it carries about that argument. In addition to arguing that the model is empirically adequate, I explore its application to two problems. The first concerns a linguistic question: why some transitive verbs permit implicit direct objects ("John ate Ø") and others do not ("*John brought Ø"). It has often been observed informally that the omission of objects is connected to the ease with which the object can be inferred. I have made this observation more formal by positing a relationship between selectional constraints and inferability. This predicts (i) that verbs permitting implicit objects select more strongly for (i.e. carry more information about) that argument than verbs that do not, and (ii) that strength of selection is a predictor of how often verbs omit their objects in naturally occurring utterances. Computational experiments confirm these predictions. Second, I have explored the practical applications of the model in resolving syntactic ambiguity. A number of authors have recently begun investigating the use of corpus-based lexical statistics in automatic parsing; the results of computational experiments using the present model suggest that many lexical relationships are better viewed in terms of underlying conceptual relationships. Thus the information-theoretic measures proposed here can serve not only as components in a theory of selectional constraints, but also as tools for practical natural language processing. Philip Stuart Resnik http://repository.upenn.edu/ircs_reports/199 http://repository.upenn.edu/ircs_reports/199 Wed, 20 Sep 2006 07:51:46 PDT Much planning research assumes that the goals for which one plans are known in advance. That is not true of trauma management, which involves both a search for relevant goals and reasoning about how to achieve them. TraumAID is a consultation system for the diagnosis and treatment of multiple trauma. It has been under development jointly at the University of Pennsylvania and the Medical College of Pennsylvania for the past eight years. TraumAID integrates diagnostic reasoning, planning and action. Its reasoner identifies diagnostic and therapeutic goals appropriate to the physician's knowledge of the patient's state, while its planner advises on beneficial actions to next perform. The physician's lack of complete knowledge of the situation and the time limitations of emergency medicine constrain the ability of any planner to identify what would be the best thing to do. Nevertheless, TraumAID's Progressive Horizon Planner has been designed to create a plan for patient care that is in keeping with the standards of managing trauma. Ron Rymon http://repository.upenn.edu/ircs_reports/198 http://repository.upenn.edu/ircs_reports/198 Wed, 20 Sep 2006 07:42:32 PDT Human agents are extremely flexible in dealing with Natural Language instructions. I argue that most instructions don't exactly mirror the agent's knowledge, but are understood by accommodating them in the context of the general plan the agent is considering: the accommodation process is guided by the goal(s) that the agent is trying to achieve. Therefore a NL system which interprets instructions must be able to recognize and/or hypothesize goals; it must make use of a flexible knowledge representation system, able to support the specialized inferences necessary to deal with input action descriptions that do not exactly match the stored knowledge. The data that support my claim are Purpose Clauses (PCs), infinitival constructions as in Do α to do β, and Negative Imperatives. I present a pragmatic analysis of both PCs and Negative Imperatives. Furthermore, I analyze the computational consequences of PCs, in terms of the relations between actions PCs express, and of the inferences an agent has to perform to understand PCs. I propose an action representation formalism that provides the required flexibility. It has two components. The Terminological Box (TBox) encodes linguistic knowledge about actions, and is expressed by means of the hybrid system CLASSIC [Brachman et al., 1991]. To guarantee that the primitives of the representation are linguistically motivated, I derive them from Jackendoff's work on Conceptual Structures [1983; 1990]. The Action Library encodes planning knowledge about actions. The action terms used in the plans are those defined in the TBox. Finally, I present an algorithm that implements inferences necessary to understand Do α to do β, and supported by the formalism I propose. In particular, I show how the TBox classifier is used to infer whether α can be assumed to match one of the substeps in the plan for β, and how expectations necessary for the match to hold are computed. Barbara Di Eugenio http://repository.upenn.edu/ircs_reports/197 http://repository.upenn.edu/ircs_reports/197 Wed, 20 Sep 2006 07:23:10 PDT We describe a density-adaptive reinforcement learning and a density-adaptive forgetting algorithm. This learning algorithm uses hybrid k-D/2k-trees to allow for a variable resolution partitioning and labelling of the input space. The density adaptive forgetting algorithm deletes observations from the learning set depending on whether subsequent evidence is available in a local region of the parameter space. The algorithms are demonstrated in a simulation for learning feasible robotic grasp approach directions and orientations and then adapting to subsequent mechanical failures in the gripper. Marcos Salganicoff http://repository.upenn.edu/ircs_reports/196 http://repository.upenn.edu/ircs_reports/196 Wed, 20 Sep 2006 07:18:15 PDT Memory-based learning algorithms lack a mechanism for tracking time-varying associative mappings. To widen their applicability, they must incorporate explicit forgetting algorithms to selectively delete observations. We describe Time-Weighted, Locally-Weighted and Performance-Error Weighted forgetting algorithms. These were evaluated with a Nearest-Neighbor Learner in a simple classification task. Locally-Weighted Forgetting outperformed Time-Weighted Forgetting under time-varying sampling distributions and mappings, and did equally well when only the mapping varied. Performance-Error forgetting tracked about as well as the other algorithms, but was superior since it permitted the Nearest-Neighbor learner to approach the Bayes' misclassification rate when the input-output mapping became stationary. Marcos Salganicoff http://repository.upenn.edu/ircs_reports/195 http://repository.upenn.edu/ircs_reports/195 Wed, 20 Sep 2006 07:11:48 PDT This paper describes a method for robotic manipulation that uses direct image-space calculation of optical flow information for continuous real-time control of manipulative actions. State variables derived from optical flow measurements are described. The resulting approach is advantageous since it robustifies the system to changes in optical parameters and also simplifies the implementation needed to succeed in the task execution. Two reference tasks and their corresponding experiments are described: the insertion of a pen into a "cap" (the capping experiment) and the rotational point-contact pushing of an object of unknown shape, mass and friction to a specified goal point in the image-space. Marcos Salganicoff http://repository.upenn.edu/ircs_reports/194 http://repository.upenn.edu/ircs_reports/194 Wed, 20 Sep 2006 07:06:39 PDT We describe an unsupervised on-line method for learning of manipulative actions that allows a robot to push an object connected to it with a rotational point contact to a desired point in image-space. By observing the results of its actions on the object's orientation in image-space, the system forms a predictive forward empirical model. This acquired model is used on-line for manipulation planning and control as it improves. Rather than explicitly inverting the forward model to achieve trajectory control, a stochastic action selection technique [Moore, 1990] is used to select the most informative and promising actions, thereby integrating active perception and learning by combining on-line improvement, task-directed exploration, and model exploitation. Simulation and experimental results of the approach are presented. Marcos Salganicoff http://repository.upenn.edu/ircs_reports/193 http://repository.upenn.edu/ircs_reports/193 Tue, 19 Sep 2006 13:58:42 PDT This thesis investigates the relationship between language behavior and agents' resource bounds by examining the use of INFORMATIONALLY REDUNDANT UTTERANCES (IRUs) in problem-solving dialogues. The content of an IRU is a proposition that the conversants already know or could infer. Since communication is a subcase of action, the existence of IRUs is a paradox because IRUs appear to be actions whose effects have already been achieved. The explication of the paradox of IRUs has ramifications for models of action in general and of dialogue in particular. I argue that IRUs can only be explained by a processing model of dialogue that reflects agents' autonomy and limited attentional and inferential capacity. The central thesis is that the communicative function of IRUs is related to the cognitive properties of resource-limited agents. In order to investigate this interaction the thesis relies on two empirical methods: (1) distributional analysis of IRUs in a large corpus of naturally occurring dialogues, and (2) dialogue simulations in the Design-World environment which supports the parameterization of the dialogue situation, the task definition, and agents' cognitive properties. The distributional analysis provides support for the claimed communicative functions by showing that IRUs demonstrate agents' autonomy and are used to support deliberation and inference. IRUs help autonomous agents coordinate on a collaborative task given their resource limits. The Design-World simulations show that discourse strategies that include IRUs are beneficial when agents are attention and inference limited or when the task is fault intolerant, inferentially complex, or requires a high degree of agreement. While some types of IRUs are beneficial simply as a rehearsal strategy, the general result is that IRUs are most beneficial when targeted at specific requirements of the communication situation. Marilyn A. Walker