AutoTutor: Incorporating Back-Channel Feedback and Other Human-Like Conversational Behaviors into an Intelligent Tutoring System

This paper describes our recent attempts to incorporate human-like conversational behaviors into the dialog moves delivered by an animated pedagogical agent that simulates human tutors. We first present a brief overview of the modules comprising AutoTutor, an intelligent tutoring system. The second section describes a set of conversational behaviors that are being incorporated into AutoTutor. The behaviors of interest involve variations in intonation, head movements, arm and hand movements, facial expressions, eye blinking, gaze direction, and back-channel feedback. The final section presents a recent empirical study concerned with back-channel feedback events during human-to-human tutoring sessions. The back-channel feedback events emitted by tutors are mostly positive (63%), mostly verbal (77%), and immediately follow speech-act boundaries or noun-phrase boundaries (83%). Tutors also deliver back-channelevents at a very high rate when students are emitting dialog, about 13 events per minute. Conversely, 88% of students' back-channel feedback events are head nods, and they occur at unbounded locations (63%).     

Metrics for shot boundary detection in digital video sequences

The detection of shot boundaries in video sequences is an important task for generating indexed video databases. This paper provides a comprehensive quantitative comparison of the metrics that have been applied to shot boundary detection. In addition, several standardized statistical tests that have not been applied to this problem, as well as three new metrics, are considered. A mathematical framework for quantitatively comparing metrics is supplied. Experimental results based on a video database containing 39,000 frames are included.     

Optimization of shell buckling incorporating Karhunen-Loève-based geometrical imperfections

The optimization of shell buckling is performed considering peak normal force and absorbed internal energy in the presence of geometrical imperfections implemented through Karhunen-Loève expansions. Initially, the mass of a shell is minimized in the presence of random initial imperfections by allowing cutouts in the material, subject to constraints on the average peak force and average internal energy. Then, robustness is considered by minimizing the coefficient of variation of the normal peak force while constraining the average peak force and average internal energy. LS-OPT® is used both to generate an experimental design and to perform a Monte Carlo simulation (96 runs) using LS-DYNA® at each of the experimental design points. The effect of imperfections when minimizing the mass is not large, but when considering robustness, however, the optimal design has a substantially increased hole size and increased shell thickness, resulting in a heavier design with maximal robustness within the constraints.     

Optical information for car following: the driving by visual angle (DVA) model

OBJECTIVE: The present study developed and tested a model of car following by human drivers. BACKGROUND: Previous models of car following are based on 3-D parameters such as lead vehicle speed and distance information, which are not directly available to a driver. In the present paper we present the driving by visual angle (DVA) model, which is based on the visual information (visual angle and rate of change of visual angle) available to the driver. METHOD: Two experiments in a driving simulator examined car-following performance in response to speed variations of a lead vehicle defined by a sum of sine wave oscillations and ramp acceleration functions. In addition, the model was applied to six driving events using real world-driving data. RESULTS: The model provided a good fit to car-following performance in the driving simulation studies as well as in real-world driving performance. A comparison with the advanced interactive microscopic simulator for urban and nonurban networks (AIMSUN) model, which is based on 3-D parameters, suggests that the DVA was more predictive of driver behavior in matching lead vehicle speed and distance headway. CONCLUSION: Car-following behavior can be modeled using only visual information to the driver and can produce performance more predictive of driver performance than models based on 3-D (speed or distance) information. APPLICATION: The DVA model has applications to several traffic safety issues, including automated driving systems and traffic flow models.     

Novel system for coupling to surface-plasmon polaritons

A novel system for coupling to surface-plasmon polaritons has been designed, fabricated, and assembled. Microcomputer controlled electronics provide for accurate and repeatable angular positioning of the waveguide coupler with respect to the incident beam and for precise angular registration of detector output signals. The optical components, combined with a retroreflecting spherical-surface coupling prism, give apertured control of the beam profile at the coupling interface.     

Effect of argon pressure on the optical properties of sputtered solar selective surfaces

The influence of argon gas pressure (0.15-40 Pa) on the refractive indices n,k of dc planar-magnetron reactively sputtered cermet and amorphous semiconductor films has been investigated for a layer thickness of approximately 50 nm. Stainless steel-carbon and amorphous hydrogenated carbon layers with relatively low index n and stainless steel-silicon and amorphous hydrogenated silicon layers with relatively high index n are examined with a view to solar selective surface applications. The development of structural porosity with associated reduction in n for layers deposited at high argon pressure significantly improves the solar absorptance of surfaces incorporating stainless steel-silicon or amorphous hydrogenated silicon layers.     

Compact yet efficient hardware implementation of artificial neural networks with customized topology

There are several neural network implementations using either software, hardware-based or a hardware/software co-design. This work proposes a hardware architecture to implement an artificial neural network (ANN), whose topology is the multilayer perceptron (MLP). In this paper, we explore the parallelism of neural networks and allow on-the-fly changes of the number of inputs, number of layers and number of neurons per layer of the net. This reconfigurability characteristic permits that any application of ANNs may be implemented using the proposed hardware. In order to reduce the processing time that is spent in arithmetic computation, a real number is represented using a fraction of integers. In this way, the arithmetic is limited to integer operations, performed by fast combinational circuits. A simple state machine is required to control sums and products of fractions. Sigmoid is used as the activation function in the proposed implementation. It is approximated by polynomials, whose underlying computation requires only sums and products. A theorem is introduced and proven so as to cover the arithmetic strategy of the computation of the activation function. Thus, the arithmetic circuitry used to implement the neuron weighted sum is reused for computing the sigmoid. This resource sharing decreased drastically the total area of the system. After modeling and simulation for functionality validation, the proposed architecture synthesized using reconfigurable hardware. The results are promising.     

Preference-based multi-objective evolutionary algorithms for power-aware application mapping on NoC platforms

Network-on-chip (NoC) are considered the next generation of communication infrastructure in embedded systems. In the platform-based design methodology, an application is implemented by a set of collaborative intellectual property (IP) blocks. The selection of the most suited set of IPs as well as their physical mapping onto the NoC infrastructure to implement efficiently the application at hand are two hard combinatorial problems that occur during the synthesis process of Noc-based embedded system implementation. In this paper, we propose an innovative preference-based multi-objective evolutionary methodology to perform the assignment and mapping stages. We use one of the well-known and efficient multi-objective evolutionary algorithms NSGA-II and microGA as a kernel. The optimization processes of assignment and mapping are both driven by the minimization of the required silicon area and imposed execution time of the application, considering that the decision maker’s preference is a pre-specified value of the overall power consumption of the implementation.     

Fractional order and BICO disturbance observers for a run-of-mine ore milling circuit

Grinding mill circuits are hard to control due to poor plant models, large external disturbances, uncertainties from internal couplings, and process variables that are difficult to measure. This paper proposes a novel fractional order disturbance observer (FO-DOB) for a run-of-mine (ROM) ore milling circuit. A fractional order low pass filter (Q-filer) is used in the DOB to offer an additional degree of freedom in tuning for set-point tracking performance and disturbance rejection performance. Another disturbance observer is introduced in which a Bode ideal cut-off (BICO) filter is used for the Q-filter. A full non-linear plant model is used for evaluation of the performance gained over the ubiquitous PI controller. The simulation results show that the FO-DOB and BICO-DOB schemes are useful additional tools for ROM ore milling circuit control implementations.     

Recognition of depressed affect in hospitalized psychiatric patients: staff and patient perceptions

A comparison of self-report vs. observer rating of depressed mood in a heterogenous inpatient population revealed wide variations in concordance among diagnostic groups. Patients diagnosed as having Affective Psychosis and "Other' illnesses showed the highest correlation between four self-report scales and an observer rating scale. Patients with a diagnosis of depressive Neurosis showed only modest correlation, while Schizophrenics revealed no significant correlation, on these instruments, suggesting inconsistent communication of affect from Schizophrenic patients to observers. In contrast, when self-report scales were intercorrelated, patients in all four diagnostic categories showed highly significant correlations, indicating that they were consistently reporting their affective state on these instruments. The implications of these findings for future research as well as for practical clinical management are discussed.