A comparative evaluation of interactive segmentation algorithms

In this paper we present a comparative evaluation of four popular interactive segmentation algorithms. The evaluation was carried out as a series of user-experiments, in which participants were tasked with extracting 100 objects from a common dataset: 25 with each algorithm, constrained within a time limit of 2 min for each object. To facilitate the experiments, a “scribble-driven” segmentation tool was developed to enable interactive image segmentation by simply marking areas of foreground and background with the mouse. As the participants refined and improved their respective segmentations, the corresponding updated segmentation mask was stored along with the elapsed time. We then collected and evaluated each recorded mask against a manually segmented ground truth, thus allowing us to gauge segmentation accuracy over time. Two benchmarks were used for the evaluation: the well-known Jaccard index for measuring object accuracy, and a new fuzzy metric, proposed in this paper, designed for measuring boundary accuracy. Analysis of the experimental results demonstrates the effectiveness of the suggested measures and provides valuable insights into the performance and characteristics of the evaluated algorithms.     

Flow map layout via spiral trees

Flow maps are thematic maps that visualize the movement of objects, such as people or goods, between geographic regions. One or more sources are connected to several targets by lines whose thickness corresponds to the amount of flow between a source and a target. Good flow maps reduce visual clutter by merging (bundling) lines smoothly and by avoiding self-intersections. Most flow maps are still drawn by hand and only few automated methods exist. Some of the known algorithms do not support edge-bundling and those that do, cannot guarantee crossing-free flows. We present a new algorithmic method that uses edge-bundling and computes crossing-free flows of high visual quality. Our method is based on so-called spiral trees, a novel type of Steiner tree which uses logarithmic spirals. Spiral trees naturally induce a clustering on the targets and smoothly bundle lines. Our flows can also avoid obstacles, such as map features, region outlines, or even the targets. We demonstrate our approach with extensive experiments.     

Marking student programs using graph similarity

We present a novel approach to the automated marking of student programming assignments. Our technique quantifies the structural similarity between unmarked student submissions and marked solutions, and is the basis by which we assign marks. This is accomplished through an efficient novel graph similarity measure (AssignSim). Our experiments show good correlation of assigned marks with that of a human marker.     

Psychometric properties of the Problematic Internet Use Questionnaire

Several instruments have been designed to measure problems associated with excessive, compulsive, or addictive use of the Internet. One such instrument, the 18-item Problematic Internet Use Questionnaire, was recently published with data supporting a three subscale model (Demetrovics et al., 2008). These researches utilized an online format with a sample taken from the general population of Hungary. We utilized an American college student sample and a paper and pencil format to perform a confirmatory factor analysis of the PIUQ. In addition, we examined the reliability and construct validity of the PIUQ by examining the scales’ relationship with several indices of psychological and physical health. CFA results indicate a barely adequate and not completely problem free three factor model for the PIUQ (χ² = 477.40; root mean square error = .097; comparative fit index = .831; Tucker Lewis coefficient = .804). Cronbach’s α for the total scale was .91 while the Cronbach’s α for each subscale were .81, .77, and .79. Construct validity for the model is demonstrated with significant correlations between the subscales and several indices of psychological and physical health. Suggestions for further research are provided.     

Emerging AI & Law approaches to automating analysis and retrieval of electronically stored information in discovery proceedings

This article provides an overview of, and thematic justification for, the special issue of the journal of Artificial Intelligence and Law entitled “E-Discovery”. In attempting to define a characteristic “AI & Law” approach to e-discovery, and since a central theme of AI & Law involves computationally modeling legal knowledge, reasoning and decision making, we focus on the theme of representing and reasoning with litigators’ theories or hypotheses about document relevance through a variety of techniques including machine learning. We also identify two emerging techniques for enabling users’ document queries to better express the theories of relevance and connect them to documents: social network analysis and a hypothesis ontology.     

A rapid computational filter for predicting the rate of human renal clearance

In silico models that predict the rate of human renal clearance for a diverse set of drugs, that exhibit both active secretion and net re-absorption, have been produced using three statistical approaches. Partial Least Squares (PLS) and Random Forests (RF) have been used to produce continuous models whereas Classification And Regression Trees (CART) has only been used for a classification model. The best models generated from either PLS or RF produce significant models that can predict acids/zwitterions, bases and neutrals with approximate average fold errors of 3, 3 and 4, respectively, for an independent test set that covers oral drug-like property space. These models contain additional information on top of any influence arising from plasma protein binding on the rate of renal clearance. Classification And Regression Trees (CART) has been used to generate a classification tree leading to a simple set of Renal Clearance Rules (RCR) that can be applied to man. The rules are influenced by lipophilicity and ion class and can correctly predict 60% of an independent test set. These percentages increase to 71% and 79% for drugs with renal clearances of < 0.1 ml/min/kg and > 1 ml/min/kg, respectively. As far as the authors are aware these are the first set of models to appear in the literature that predict the rate of human renal clearance and can be used to manipulate molecular properties leading to new drugs that are less likely to fail due to renal clearance.     

<Emphasis Type="Italic">A Midsummer Night’s Dream</Emphasis> (with flying robots)

Seven flying robot “fairies” joined human actors in the Texas A&;M production of William Shakespeare’s A Midsummer Night’s Dream. The production was a collaboration between the departments of Computer Science and Engineering, Electrical and Computer Engineering, and Theater Arts. The collaboration was motivated by two assertions. First, that the performing arts have principles for creating believable agents that will transfer to robots. Second, the theater is a natural testbed for evaluating the response of untrained human groups (both actors and the audience) to robots interacting with humans in shared spaces, i.e., were believable agents created? The production used two types of unmanned aerial vehicles, an AirRobot 100-b quadrotor platform about the size of a large pizza pan, and six E-flite Blade MCX palm-sized toy helicopters. The robots were used as alter egos for fairies in the play; the robots did not replace any actors, instead they were paired with them. The insertion of robots into the production was not widely advertised so the audience was the typical theatergoing demographic, not one consisting of people solely interested technology. The use of radio-controlled unmanned aerial vehicles provides insights into what types of autonomy are needed to create appropriate affective interactions with untrained human groups. The observations from the four weeks of practice and eight performances contribute (1) a taxonomy and methods for creating affect exchanges between robots and untrained human groups, (2) the importance of improvisation within robot theater, (3) insights into how untrained human groups form expectations about robots, and (4) awareness of the importance of safety and reliability as a design constraint for public engagement with robot platforms. The taxonomy captures that apparent affect can be created without explicit affective behaviors by the robot, but requires talented actors to convey the situation or express reactions. The audience’s response to robot crashes was a function of whether they had the opportunity to observe how the actors reacted to robot crashes on stage, suggesting that pre-existing expectations must be taken into account in the design of autonomy. Furthermore, it appears that the public expect robots to be more reliable (an expectation of consumer product hardening) and safe (an expectation from product liability) than the current capabilities and this may be a major challenge or even legal barrier for introducing robots into shared public spaces. These contributions are expected to inform design strategies for increasing public engagement with robot platforms through affect, and shows the value of arts-based approaches to public encounters with robots both for generating design strategies and for evaluation.     

Effects of soil moisture and water depth on ERS SAR backscatter measurements from an Alaskan wetland complex

We conducted a preliminary investigation of the response of ERS C-band SAR backscatter to variations in soil moisture and surface inundation in wetlands of interior Alaska. Data were collected from 5 wetlands over a three-week period in 2007. Results showed a positive correlation between backscatter and soil moisture in sites dominated by herbaceous vegetation cover (r = 0.74, p < 0.04). ERS SAR backscatter was negatively correlated to water depth in all open (non-forested) wetlands when water table levels were more than 6 cm above the wetland surface (r = − 0.82, p < 0.001). There was no relationship between backscatter and soil moisture in the forested (black spruce-dominated) wetland site. Our preliminary results show that ERS SAR data can be used to monitor variations in hydrologic conditions in high northern latitude wetlands (including peatlands), particularly sites with sparse tree cover.     

User-input driven QoS management in ad hoc networks

Though the network quality-of-service (QoS) metrics are defined in terms of technical parameters (e.g., delay, jitter, bandwidth), they are rather subjective when it comes to the end user. Oftentimes, the end user finds it difficult to express his desired QoS in such technical parameters, though he has a fair idea of what QoS he desires. In this paper, we show how translation functions can be devised and used to translate the user inputs to networking parameters that are used by various layers of the protocol stack. In particular, we consider an ad hoc network and show an interface design that uses translation functions to map user supplied inputs to parameters at the medium access control (MAC) and routing layers. These parameters, in turn, choose the right strategies that particular layer functionality can adopt, such that the QoS desired by the user is achieved. We also compute the associated costs due to the different strategies adopted. We implement the interface on ns-2 and conduct simulation experiments with randomly scattered IEEE 802.11 enabled nodes. Results show the functionality of the interface and demonstrate how delay, throughput, and network lifetime are affected when the end user seeks different levels of QoS.     

The design and implementation of a session layer for delay-tolerant networks

The DTN architecture is based around sender-initiated unicast communication that is insufficient or inconvenient to meet the needs of many applications. To address these limitations, we define a DTN session layer and associated extensions to the DTN bundle protocol that more naturally support receiver-driven applications and multicast communication. Within a session, we provide mechanisms allowing applications to convey ordering relationships between successive transmissions that can be used by the network to help ensure a distributed application’s delivery ordering expectations are met. We also extend the bundle protocol’s expiration procedures to support more efficient network utilization by allowing in-network deletion of obsolete messages. We present the design rationale and describe our implementation of these mechanisms and discuss their advantages in meeting the needs of several popular types of applications.