Papers by Srini Narayanan
Computer Vision – ECCV 2020 Workshops, 2020
We propose a lightweight real-time sign language detection model, as we identify the need for suc... more We propose a lightweight real-time sign language detection model, as we identify the need for such a case in videoconferencing. We extract optical flow features based on human pose estimation and, using a linear classifier, show these features are meaningful with an accuracy of 80%, evaluated on the DGS Corpus. Using a recurrent model directly on the input, we see improvements of up to 91% accuracy, while still working under 4ms. We describe a demo application to sign language detection in the browser in order to demonstrate its usage possibility in videoconferencing applications.
allow one to mark up web pages to indicate the meaning of their content; it is intended that the ... more allow one to mark up web pages to indicate the meaning of their content; it is intended that the results delivered by a DAML-enabled browser will more closely match the intentions of the user than is possible with today’s syntactically oriented search engines. In this paper we present our vision of a DAML-enabled search architecture. We present a set of queries of increasing complexity that should be answered efficiently in a Semantic Web. We describe several scenarios illustrating how queries are processed, identifying the main software components necessary to facilitate the search. We examine the issue of inference in search, and we address how to characterize procedures and services in DAML, enabling a DAML query language to find web sites with specified capabilities.
The Semantic Web should enable greater access not only to content but also to services on the Web... more The Semantic Web should enable greater access not only to content but also to services on the Web. Users and software agents should be able to discover, invoke, compose, and monitor Web resources offering particular services and having particular properties. As part of the DARPA Agent Markup Language program, we have begun to develop DAML-S, a DAML+OIL ontology for describing the properties and capabilities of web services, and that supports the automatic discovery, invocation, composition and monitoring of these services. In this paper we describe the overall structure of the ontology, the service profile for advertising services, and the process model for the detailed description of the operation of services. We also compare DAML-S with several industry efforts to define standards for characterizing services on the Web. 1 Introduction: Services
The Semantic Web — ISWC 2002, 2002
In this paper we present DAML-S, a DAML+OIL ontology for describing the properties and capabiliti... more In this paper we present DAML-S, a DAML+OIL ontology for describing the properties and capabilities of Web Services. Web Services-Web-accessible programs and devices-are garnering a great deal of interest from industry, and standards are emerging for low-level descriptions of Web Services. DAML-S complements this effort by providing Web Service descriptions at the application layer, describing what a service can do, and not just how it does it. In this paper we describe three aspects of our ontology: the service profile, the process model, and the service grounding. The paper focuses on the grounding, which connects our ontology with low-level XML-based descriptions of Web Services.
Computational Linguistics, 2017
Highly frequent in language and communication, metaphor represents a significant challenge for Na... more Highly frequent in language and communication, metaphor represents a significant challenge for Natural Language Processing (NLP) applications. Computational work on metaphor has traditionally evolved around the use of hand-coded knowledge, making the systems hard to scale. Recent years have witnessed a rise in statistical approaches to metaphor processing. However, these approaches often require extensive human annotation effort and are predominantly evaluated within a limited domain. In contrast, we experiment with weakly supervised and unsupervised techniques—with little or no annotation—to generalize higher-level mechanisms of metaphor from distributional properties of concepts. We investigate different levels and types of supervision (learning from linguistic examples vs. learning from a given set of metaphorical mappings vs. learning without annotation) in flat and hierarchical, unconstrained and constrained clustering settings. Our aim is to identify the optimal type of superv...
Conceptual metaphor is a widespread phenomenon in language and thought. While there has been prog... more Conceptual metaphor is a widespread phenomenon in language and thought. While there has been progress in the computational modeling of metaphoric inference (Narayanan1997,99), there is no existing model model of metaphor acquisition that matches psycholinguistic data (Johnson 1997). This paper suggests that the combination of Spike Timing Dependent Plasticity (STDP) (a widely prevalent learning mechanism found in many areas of the brain) and the specific dynamics of the input activation patterns culled from the psycholinguistic data, can provide a neurally plausible account of metaphor acquisition. The dynamics of the interaction between the learning mechanism and the developmental input sequence seems also to provide insights on puzzling paradoxes in the dynamics of metaphor learning, including the partial nature of the mappings. The theory presented is based on a computational simulation of STDP with parameters culled from the experimental literature. As far as we are aware, the w...
This abstract describes a computational model that cashes out the belief that metaphor interpreta... more This abstract describes a computational model that cashes out the belief that metaphor interpretation is grounded in embodied primitives. The specific task addressed is the interpretation of simple causal narratives in the domains of Politics and Economics. The stories are taken from newspaper articles in these domains. When presented with a pre-parsed version of these narratives as input, the system described is able to generate commonsense inferences consistent with the input. Recent work in Cognitive Semantics (LJ 1980; Lakoff 1994; Talmy 1985; 1987) suggests that the structure abstract actions (such as states, causes, purposes, means, etc.) are characterized cognitively in terms of embodied concepts from the domains of force, motion, and space. Such observations have resulted in proposals that abstract reason is grounded in recurring patterns of experience called/mate Schemas (Lakoff 1987). However, the work Cognitive Semantics lacks any computational model for such theories, an...
We present a connectionist realization of parameterized schemas that can model highlevel sensory-... more We present a connectionist realization of parameterized schemas that can model highlevel sensory-motor processes and be a candidate representation for implementing reactive behaviors. The connectionist realization involves a number of ideas including the use of focal-clusters and feedback loops to control a distributed process without a central controller and the expression and propagation of dynamic bindings via temporal synchrony. We employ a uniform mechanism for interaction between schemas, low-level somatosensory and proprioceptive processes, and high-level reasoning and memory processes. Our representation relates to work in connectionist models of rapid-reflexive-reasoning and also suggests solutions to several problems in language acquisition and understanding.
The process of human blood clotting involves a complex interaction of continuous-time/continuous-... more The process of human blood clotting involves a complex interaction of continuous-time/continuous-state processes and discrete-event/discrete-state phenomena, where the former comprise the various chemical rate equations (which can be written as differential equations) and the latter comprise both threshold-limited behaviors and qualitative facets of the coagulation cascade. We model this process as a hybrid dynamical system, consisting of both discrete and continuous dynamics. Previous blood clotting models used only continuous dynamics and perforce addressed only portions of the coagulation cascade. The model was implemented as a hybrid Petri net, a graphical modeling language that extends ordinary Petri nets to cover continuous quantities and continuous-time flows. The primary focus is simulation; specifically, we are interested in (1) fidelity of simulation to the actual clotting process in terms of clotting factor concentrations and elapsed time; (2) simulating deficiencies, surfeits, and other perturbations of initial values of blood proteins, and their consequences for clotting factor concentrations and for clotting time, especially in the context of known clotting pathologies; and (3) providing fine-grained predictions which may be used to refine clinical understanding of blood clotting. * Our current implementation is based on the Visual Object Net++ platform, a dedicated HPN modeling and simulation environment [9], which includes a graphical language that offers a suite of object-oriented programming (OOP) features: hierarchical organization, inheritance and object reuse. * Of course, no tissue factor is actually present pre-injury; this quantity represents the amount released immediately upon blood vessel rupture †These are the unactivated and activated concentrations, respectively, of factor VII
2010 Aaai Fall Symposium Series, Mar 11, 2010
2010 Aaai Fall Symposium Series, 2010
Abstract Narratives structure our understanding of the world and of ourselves. They exploit the s... more Abstract Narratives structure our understanding of the world and of ourselves. They exploit the shared cognitive structures of human motivations, goals, actions, events, and outcomes. We report on a computational model that is motivated by results in neural computation and ...
... schemas and converting the various place and arc types to their equivalent Petri net primitiv... more ... schemas and converting the various place and arc types to their equivalent Petri net primitives without ... In this case, through a well de ned net transformation (Murata 1989), we get aa ... To this end, we outlined a novel computational representa-tion and simulation, inspired by well ...
... Metaphor and Aspect by Srinivas Sankara Narayanan BS IIT Madras 1985 A dissertation submitted... more ... Metaphor and Aspect by Srinivas Sankara Narayanan BS IIT Madras 1985 A dissertation submitted in partial satisfaction of the requirements for the degree of ... by Srinivas Sankara Narayanan Doctor of Philosophy in Computer Science University of California, Berkeley ...
Proceedings of Frame Semantics in NLP: A Workshop in Honor of Chuck Fillmore (1929-2014), 2014
FrameNet is the best currently operational version of Chuck Fillmore's Frame Semantics. As FrameN... more FrameNet is the best currently operational version of Chuck Fillmore's Frame Semantics. As FrameNet has evolved over the years, we have been building a series of increasingly ambitious prototype systems that exploit FrameNet as a semantic resource. Results from this work point to frames as a natural representation for applications that require linking textual meaning to world knowledge.
Human language processing relies on many kinds of linguistic knowledge, and is sensitive to their... more Human language processing relies on many kinds of linguistic knowledge, and is sensitive to their frequency, including lexical frequencies (
Proceedings of the Sixteenth National Conference on Artificial Intelligence and the Eleventh Innovative Applications of Artificial Intelligence Conference Innovative Applications of Artificial Intelligence, 1999
... What is Metaphor?. Advances in Connec-tionist Theory. 1/3 : Analogical Connections, V3,1994. ... more ... What is Metaphor?. Advances in Connec-tionist Theory. 1/3 : Analogical Connections, V3,1994. Martin, J. (1990). A Computational Model of Metaphor In-terpretation. ... Pearson KG (1993). Common Principles of Motor Control in Vertebrates and Invertebrates. Ann. ...
Uploads
Papers by Srini Narayanan