Dealing with multiple experts and non-stationarity in inverse reinforcement learning: an application to real-life problems

Machine Learning - In real-world applications, inferring the intentions of expert agents (e.g., human operators) can be fundamental to understand how possibly conflicting objectives are managed,...     

Dynamics and control of multiple cooperating manipulators with rolling contacts

In this article, we investigate the control of multiple cooperating mechanisms manipulating a common object, with rolling contact between the end-effectors and the object. We derive new kinematic and dynamic formulations for the system, in which the rolling degrees of freedom between the end-effector and the object are modeled as one or more unactuated joints of the manipulator. Differential geometric concepts are used to interpret the curvature and torsion forms used in contact kinematic equations derived by Montana. The augmentation of the rolling contacts imparts each manipulator with additional degrees of freedom and could also make it kinematically redundant. This can be exploited for the satisfaction of a secondary subtask criterion, such as collision avoidance, which is demonstrated in the examples. In addition, a control law can be specified that enables simultaneous control of the object trajectory, the internal forces in the object, and the trajectories of the contact points on the object or the end-effectors. © 1996 John Wiley & Sons, Inc.     

Dealing with Prosody in a Text-to-Speech System

The task of assigning appropriate intonation to syntheticspeech is one that requires knowledge of linguistic structure as well ascomputational possibilities. This paper surveys the basic challengesfacing the designer of a text-to-speech system, and reviews some of theperspectives on these problems that have been developed in thelinguistic literature.     

Teaching robots to cooperate with humans in dynamic manipulation tasks based on multi-modal human-in-the-loop approach

We propose an approach to efficiently teach robots how to perform dynamic manipulation tasks in cooperation with a human partner. The approach utilises human sensorimotor learning ability where the human tutor controls the robot through a multi-modal interface to make it perform the desired task. During the tutoring, the robot simultaneously learns the action policy of the tutor and through time gains full autonomy. We demonstrate our approach by an experiment where we taught a robot how to perform a wood sawing task with a human partner using a two-person cross-cut saw. The challenge of this experiment is that it requires precise coordination of the robot’s motion and compliance according to the partner’s actions. To transfer the sawing skill from the tutor to the robot we used Locally Weighted Regression for trajectory generalisation, and adaptive oscillators for adaptation of the robot to the partner’s motion.     

Dealing with Satisfiability and n-ary CSPs in a Logical Framework

We investigate in this work a generalization of the known CNF representation that we call General Normal Form (GNF) and extend the resolution rule of Robinson to design a new sound and complete resolution proof system for the GNF. We show that by using a cardinality operator in the GNF we obtain the new representation CGNF that allows a natural and efficient Boolean encoding for n-ary finite discrete extensional CSPs. We prove that the space complexity of a CSP encoded in the CGNF is identical to its space complexity when it is expressed in the classical CSP representation. We prove that the generalized resolution rule applies for the CGNF encoding and introduce a new inference rule whose application until saturation achieves arc-consistency in a linear time complexity for n-ary CSP expressed in the CGNF encoding. Two enumerative methods for solving CSP instances encoded in this Boolean framework are studied: the first one (equivalent to MAC in CSPs) maintains full arc-consistency at each node of the search tree while the second (equivalent to FC in CSPs) performs partial arc-consistency on each node. Both methods are experimented and compared on some instances of the Ramsey problem, randomly generated 3/4-ary CSPs and promising results are obtained. We also adapted the notion of clause learning defined in SAT for the CGNF encoding. Finally, we show how the proposed inference rule can be used to achieve Path-consistency in the case of binary CSPs encoded in CGNF, and how it can be used to enforce strong path consistency when it is combined with the generalized resolution rule.     

Approximation of multiple integrals over hyperboloids with application to a quadratic portfolio with options

An application involving a financial quadratic portfolio, where the joint underlying log-returns follow a multivariate elliptic distribution, is considered. This motivates the need for methods for the approximation of multiple integrals over hyperboloids. Transformations are used to reduce the hyperboloid integrals to products of integrals which can be approximated with appropriate numerical methods. The application of these methods is demonstrated using some financial applications examples.     

Dealing with obsolete parts

The US Air Force has launched an initiative to develop an affordable, systematic approach to dealing with obsolete electronic parts. In addition to management and reengineering tools, the initiative is developing reliability models for commercially manufactured electronics used in defense systems     

Dealing with algebraic expressions over a field in Coq using Maple

We describe an interface between the Coq proof assistant and the Maple symbolic computation system, which mainly consists in importing, in Coq, Maple computations regarding algebraic expressions over fields. These can either be pure computations, which do not require any validation, or computations used during proofs, which must be proved (to be correct) within Coq. These correctness proofs are completed automatically thanks to the tactic Field, which deals with equalities over fields. This tactic, which may generate side conditions (regarding the denominators) that must be proved by the user, has been implemented in a reflexive way, which ensures both efficiency and certification. The implementation of this interface is quite light and can be very easily extended to get other Maple functions (in addition to the four functions we have imported and used in the examples given here).     

初识物联网

随着网络技术的不断发展和广泛的应用,物联网作为一个新兴的事物开始进入人们的日常生活中.物联网就是通过各种信息传感设备,按照约定的协议,以实现信息的交换与以及对信息的智能化管理.本文研究内容主要针对物联网的发展、原理、特性以及物联网相关的技术.     

Dealing with multiple documents on the WWW: The role of metacognition in the formation of documents models

Drawing on the theory of documents representation (Perfetti et al., Toward a theory of documents representation. In: H. v. Oostendorp & S. R. Goldman (Eds.), The construction of mental representations during reading. Mahwah, NJ: Erlbaum, 1999), we argue that successfully dealing with multiple documents on the World Wide Web requires readers to form documents models; that is, to form a representation of contents and sources. We present a study in which we tested the assumption that the use of metacognitive strategies is crucial to the formation of documents models. A total of 100 participants with little medical knowledge were asked to conduct an Internet research on a medical topic. Participants were randomly assigned to four experimental groups that received different types of metacognitive prompts: participants either received evaluation prompts, monitoring prompts, both types of prompts, or no prompts. A control group took paper-and-pencil notes. Results showed that laypersons receiving evaluation prompts outperformed controls in terms of knowledge about sources and produced more arguments relating to the source of information when justifying credibility judgments. However, laypersons receiving evaluation prompts were not better able to indicate the source of information after Internet research than controls. In addition, laypersons receiving monitoring prompts acquired significantly more knowledge about facts, and performed slightly better on a comprehension test. It is concluded that the results underline the importance of metacognition in dealing with multiple documents.     

Dealing with complex queries in decision-support systems

In decision-making problems under uncertainty, a decision table consists of a set of attributes indicating what is the optimal decision (response) within the different scenarios defined by the attributes. We recently introduced a method to give explanations of these responses. In this paper, the method is extended. To do this, it is combined with a query system to answer expert questions about the preferred action for a given instantiation of decision table attributes. The main difficulty is to accurately answer queries associated with incomplete instantiations. Incomplete instantiations are the result of the evaluation of a partial model outputting decision tables that only include a subset of the whole problem, leading to uncertain responses. Our proposal establishes an automatic and interactive dialogue between the decision-support system and the expert to elicit information from the expert to reduce uncertainty. Typically, the process involves learning a Bayesian network structure from a relevant part of the decision table and computing some interesting conditional probabilities that are revised accordingly.     

Dealing with variability in context-aware mobile software

Mobile context-aware software pose a set of challenging requirements to developers as these applications exhibit novel features, such as handling varied sensing devices and dynamically adapting to the user’s context (e.g. his or her location), and evolve quickly according to technological advances.In this paper, we discuss how to handle variability both across different domains and during the evolution of a single application. We present a set of design structures for solving different problems related with mobility (such as location sensing, behaviour adaptation, etc.), together with the design rationale underlying them, and show how these sound micro-architectural constructs impact on variability. Our presentation is illustrated with case studies in different domains.     

Dealing with Risk in Scientific Software Development

The development of scientific software involves risk in the underlying theory, its implementation, and its use. Through a series of interviews, the authors explored how research scientists at two Canadian universities developed their software. These interviews indicated that the scientists used a set of strategies to address risk. They also suggested where the software engineering community could perform research focused on specific problems faced by scientific software developers.     

Dealing with Process Overload Attacks in UNIX

ABTRACT

A process overload attack is an attack on a shared computer system in which a user repeatedly forks new processes and hence makes the system unusable for others. The specific problem we address is seen in an academic environment where student programs create unintentional process overload attacks in UNIX systems by careless coding. Instead of rebooting the system or manually examining and killing the processes, our approach to dealing with these attacks was to build a process load monitoring tool to detect and kill these processes automatically. This paper focuses on what we learned about the behaviors of different fork bombs, how we classified them based on their self-replicating capabilities, and our experience with detecting, killing and cleaning these unwanted processes.     

Dealing with early life failures

D&T editor in chief Tim Cheng discusses the industry's struggle to screen latent defects for complex ICs. He also welcomes two new editors to the D&T editorial board.