Kristina Lerman

Distributed online localization in sensor networks using a moving target

We describe a novel method for node localization in a sensor network where there are a fraction of reference nodes with known locations. For application-specific sensor networks, we argue that it makes sense to treat localization through online distributed learning and integrate it with an application task such as target tracking. We propose distributed online algorithm in which sensor nodes use geometric constraints induced by both radio connectivity and sensing to decrease the uncertainty of their position. The sensing constraints, which are caused by a commonly sensed moving target, are usually tighter than connectivity based constraints and lead to a decrease in average localization error over time. Different sensing models, such as radial binary detection and distance-bound estimation, are considered. First, we demonstrate our approach by studying a simple scenario in which a moving beacon broadcasts its own coordinates to the nodes in its vicinity. We then generalize this to the case when instead of a beacon, there is a moving target with a-priori unknown coordinates. The algorithms presented are fully distributed and assume only local information exchange between neighboring nodes. Our results indicate that the proposed method can be used to significantly enhance the accuracy in position estimation, even when the fraction of reference nodes is small. We compare the efficiency of the distributed algorithms to the case when node positions are estimated using centralized (convex) programming. Finally, simulations using the TinyOS-Nido platform are used to study the performance in more realistic scenarios.

Using a model of social dynamics to predict popularity of news

Popularity of content in social media is unequally distributed, with some items receiving a disproportionate share of attention from users. Predicting which newly-submitted items will become popular is critically important for both companies that host social media sites and their users. Accurate and timely prediction would enable the companies to maximize revenue through differential pricing for access to content or ad placement. Prediction would also give consumers an important tool for filtering the ever-growing amount of content. Predicting popularity of content in social media, however, is challenging due to the complex interactions among content quality, how the social media site chooses to highlight content, and influence among users. While these factors make it difficult to predict popularity a priori, we show that stochastic models of user behavior on these sites allows predicting popularity based on early user reactions to new content. By incorporating aspects of the web site design, such models improve on predictions based on simply extrapolating from the early votes. We validate this claim on the social news portal Digg using a previously-developed model of social voting based on the Digg user interface.

Analysis of Dynamic Task Allocation in Multi-Robot Systems

Dynamic task allocation is an essential requirement for multi-robot systems operating in unknown dynamic environments. It allows robots to change their behavior in response to environmental changes or actions of other robots in order to improve overall system performance. Emergent coordination algorithms for task allocation that use only local sensing and no direct communication between robots are attractive because they are robust and scalable. However, a lack of formal analysis tools makes emergent coordination algorithms difficult to design. In this paper we present a mathematical model of a general dynamic task allocation mechanism. Robots using this mechanism have to choose between two types of tasks, and the goal is to achieve a desired task division in the absence of explicit communication and global knowledge. Robots estimate the state of the environment from repeated local observations and decide which task to choose based on these observations. We model the robots and observations as stochastic processes and study the dynamics of the collective behavior. Specifically, we analyze the effect that the number of observations and the choice of the decision function have on the performance of the system. The mathematical models are validated in a multi-robot multi-foraging scenario. The models predictions agree very closely with results of embodied simulations.

Social Information Processing in News Aggregation

Social media sites underscore the Webs transformation to a participatory medium in which users collaboratively create, evaluate, and distribute information. Innovations in social media have led to social information processing, a new paradigm for interacting with data. The social news aggregator Digg exploits social information processing for document recommendation and rating. Additionally, via mathematical modeling, its possible to describe how collaborative document rating emerges from the independent decisions users make. Using such a model, the author reproduces observed ratings that actual stories on Digg have received.

Using the structure of Web sites for automatic segmentation of tables

Many Web sites, especially those that dynamically generate HTML pages to display the results of a users query, present information in the form of list or tables. Current tools that allow applications to programmatically extract this information rely heavily on user input, often in the form of labeled extracted records. The sheer size and rate of growth of the Web make any solution that relies primarily on user input is infeasible in the long term. Fortunately, many Web sites contain much explicit and implicit structure, both in layout and content, that we can exploit for the purpose of information extraction. This paper describes an approach to automatic extraction and segmentation of records from Web tables. Automatic methods do not require any user input, but rely solely on the layout and content of the Web source. Our approach relies on the common structure of many Web sites, which present information as a list or a table, with a link in each entry leading to a detail page containing additional information about that item. We describe two algorithms that use redundancies in the content of table and detail pages to aid in information extraction. The first algorithm encodes additional information provided by detail pages as constraints and finds the segmentation by solving a constraint satisfaction problem. The second algorithm uses probabilistic inference to find the record segmentation. We show how each approach can exploit the web site structure in a general, domain-independent manner, and we demonstrate the effectiveness of each algorithm on a set of twelve Web sites.

Wrapper maintenance: a machine learning approach

The proliferation of online information sources has led to an increased use of wrappers for extracting data from Web sources. While most of the previous research has focused on quick and efficient generation of wrappers, the development of tools for wrapper maintenance has received less attention. This is an important research problem because Web sources often change in ways that prevent the wrappers from extracting data correctly. We present an efficient algorithm that learns structural information about data from positive examples alone. We describe how this information can be used for two wrapper maintenance applications: wrapper verification and reinduction. The wrapper verification system detects when a wrapper is not extracting correct data, usually because the Web source has changed its format. The reinduction algorithm automatically recovers from changes in the Web source by identifying data on Web pages so that a new wrapper may be generated for this source. To validate our approach, we monitored 27 wrappers over a period of a year. The verification algorithm correctly discovered 35 of the 37 wrapper changes, and made 16 mistakes, resulting in precision of 0.73 and recall of 0.95. We validated the reinduction algorithm on ten Web sources. We were able to successfully reinduce the wrappers, obtaining precision and recall values of 0.90 and 0.80 on the data extraction task.

Coalition formation for large-scale electronic markets

Coalition formation is a desirable behavior in a multiagent system, when a group of agents can perform a task more efficiently than any single agent can. Computational and communications complexity of traditional approaches to coalition formation, e.g., through negotiation, make them impractical for large systems. We propose an alternative, physics-motivated mechanism for coalition formation that treats agents as randomly moving, locally interacting entities. A new coalition may form when two agents encounter one another and it may grow when a single agent encounters it. Such agent-level behavior leads to a macroscopic model that describes how the number and distribution of coalitions change with time. We increase the generality and complexity of the model by letting the agents leave coalitions with some probability. The model is expressed mathematically as a series of differential equations. These equations have steady state solutions that describe the equilibrium distribution of coalitions. Within a context of a specific multi-agent application, we analyze and discuss the connection between the global system utility the parameters of the model.

Mathematical Model of Foraging in a Group of Robots: Effect of Interference

In multi-robot applications, such as foraging or collection tasks, interference, which results from competition for space between spatially extended robots, can significantly affect the performance of the group. We present a mathematical model of foraging in a homogeneous multi-robot system, with the goal of understanding quantitatively the effects of interference. We examine two foraging scenarios: a simplified collection task where the robots only collect objects, and a foraging task, where they find objects and deliver them to some pre-specified “home” location. In the first case we find that the overall group performance improves as the system size grows; however, interference causes this improvement to be sublinear, and as a result, each robots individual performance decreases as the group size increases. We also examine the full foraging task where robots collect objects and deliver them home. We find an optimal group size that maximizes group performance. For larger group sizes, the group performance declines. However, again due to the effects of interference, the individual robots performance is a monotonically decreasing function of the group size. We validate both models by comparing their predictions to results of sensor-based simulations in a multi-robot system and find good agreement between theory and simulations data.

Resource Allocation in the Grid Using Reinforcement Learning

In this paper we study a minimalist decentralized algorithm for resource allocation in a simplified Grid-like environment. We consider a system consisting of large number of heterogenous reinforcement learning agents that share common resources for their computational needs. There is no communication between the agents: the only information that agents receive is the (expected) completion time of a job it submitted to a particular resource and which serves as a reinforcement signal for the agent. The results of our experiments suggest that reinforcement learning can be used to improve the quality of resource allocation in large scale heterogenous system.

Semi-automatically mapping structured sources into the semantic web

Linked data continues to grow at a rapid rate, but a limitation of a lot of the data that is being published is the lack of a semantic description. There are tools, such as D2R, that allow a user to quickly convert a database into RDF, but these tools do not provide a way to easily map the data into an existing ontology. This paper presents a semi-automatic approach to map structured sources to ontologies in order to build semantic descriptions (source models). Since the precise mapping is sometimes ambiguous, we also provide a graphical user interface that allows a user to interactively refine the models. The resulting source models can then be used to convert data into RDF with respect to a given ontology or to define a SPARQL end point that can be queried with respect to an ontology. We evaluated the overall approach on a variety of sources and show that it can be used to quickly build source models with minimal user interaction.