Robust Real-Time Face Detection

This paper describes a face detection framework that is capable of processing images extremely rapidly while achieving high detection rates. There are three key contributions. The first is the introduction of a new image representation called the Integral Image which allows the features used by our detector to be computed very quickly. The second is a simple and efficient classifier which is built using the AdaBoost learning algorithm (Freund and Schapire, 1995) to select a small number of critical visual features from a very large set of potential features. The third contribution is a method for combining classifiers in a cascade which allows background regions of the image to be quickly discarded while spending more computation on promising face-like regions. A set of experiments in the domain of face detection is presented. The system yields face detection performance comparable to the best previous systems (Sung and Poggio, 1998; Rowley et al., 1998; Schneiderman and Kanade, 2000; Roth et al., 2000). Implemented on a conventional desktop, face detection proceeds at 15 frames per second.     

A Stylometric Analysis of Ya?ar Kemal’s ? nce Memed Tetralogy

We analyze four nce Memed novels of Yaar Kemal using six style markers: most frequent words, syllable counts, word type – or part of speech – information, sentence length in terms of words, word length in text, and word length in vocabulary. For analysis we divide each novel into five thousand word text blocks and count the frequencies of each style marker in these blocks. The style markers showing the best separation are most frequent words and sentence lengths. We use stepwise discriminant analysis to determine the best discriminators of each style marker. We then use these markers in cross validation based discriminant analysis. Further investigation based on multiple analysis of variance (MANOVA) reveals how the attributes of each style marker group distinguish among the volumes.     

The Valid Use of Utility in Adaptive Real-Time Systems

Real-time systems that must adapt their behavior to changes in internal and external circumstances require flexibility in their scheduling. One approach that has been advocated for achieving this flexibility is called value-based scheduling, wherein services are distinguished based upon their current utility values. The main result of this paper is that the assumptions used in the assignment of these values must be matched with the way the values are used in scheduling. The key notion for ensuring this is the scale type of the value, in the sense defined by measurement theory. There are simple tests for meaningful uses of values based on their scale types, and we apply these tests to investigate the meaningfulness requirements of some commonly-used scheduling approaches.     

More Church–Rosser Proofs

The proofs of the Church–Rosser theorems for , , and reduction in untyped -calculus are formalized in Isabelle/HOL, an implementation of Higher Order Logic in the generic theorem prover Isabelle. For -reduction, both the standard proof and Takahashi's are given and compared. All proofs are based on a general theory of commutating relations that supports an almost geometric style of reasoning about confluence diagrams.     

Uniformization based sensitivity estimation for a class of discrete-event systems

Infinitestimal Perturbation Analysis (IPA) estimators are based on particular couplings of parameteric families of discrete event systems where small changes in the parameter value, typically, cause small changes in the timing of events and, for finite horizons, the sequence of states visisted remains the same. We consider another coupling approach based on the uniformization procedure and a simple generalization of it. In our case any small change in the parameter value causes a change in the state of the system; our parameterization of trajectories keeps them highly synchronized, hence the effect of such changes can be estimated, sometimes efficiently. In this framework, we define three tupes of performance sensitivity estimators for a broad class of performance measures and with respect to a range of parameter values. Performance measures on finite deterministic horizons are considered and it is shown that they are unbiased under mild conditions. We show that for some systems the derivative estimators can be calculated from a nominal sample path of the system.     

A systems approach to chemical management by manufacturing firms in an environmentally constrained world

The regional and global environmental perturbations resulting from the effects of human economic activity on fundamental biological, chemical, and physical systems can no longer be ignored. Indeed, a complex and expanding set of statutory and regulatory responses, directed in large part at industrial and manufacturing activity, demonstrates society's increasing understanding of this process. These developments have driven the evolution of industry environmental compliance and management systems. In particular, integrated chemical management systems (ICMS), which consist of two subsystems — a data subsystem and a management subsystem, have become both more complex and more integrated into traditional business operating and management systems. Three stages in ICMS evolution are defined and described: stage I, the presystemic stage; stage II, the static system stage; and stage III, the interactive system stage.     

Enhancing the performance of an agent-based manufacturing system through learning and forecasting

Agent-based technology has been identified as an important approach for developing next generation manufacturing systems. One of the key techniques needed for implementing such advanced systems will be learning. This paper first discusses learning issues in agent-based manufacturing systems and reviews related approaches, then describes how to enhance the performance of an agent-based manufacturing system through learning from history (based on distributed case-based learning and reasoning) and learning from the future (through system forecasting simulation). Learning from history is used to enhance coordination capabilities by minimizing communication and processing overheads. Learning from the future is used to adjust promissory schedules through forecasting simulation, by taking into account the shop floor interactions, production and transportation time. Detailed learning and reasoning mechanisms are described and partial experimental results are presented.     

An operation-control scheme for authorization in computer systems

The access-control authorization scheme, which is being used for the protection of operating systems, is found to be inadequate in other areas, such as in databases and information systems. A new authorization scheme, which is a natural extension of access control, is proposed. The new scheme, which is called operation control, is shown to be superior to the accesscontrol scheme in a number of ways. In particular, it facilitates more natural and efficient representations of policies, particularly the type of complex policies that appear in information systems, it facilitates enforcement by compile-time validation due to a greater stability of authority states, and it reduces the need for revocation.This work was partially supported by Grant DAHCIS-73-G6 of the Advanced Research Project Agency of the US government. This paper is a modified version of the paper An Activator-based protection scheme, July 1976 (SOSAP-TR-25).     

Modelling structures in generic space, a condition for adaptiveness of monitoring cognitive agent

The adaptiveness of agents is one of the basic conditions for the autonomy. This paper describes an approach of adaptiveness forMonitoring Cognitive Agents based on the notion of generic spaces. This notion allows the definition of virtual generic processes so that any particular actual process is then a simple configuration of the generic process, that is to say a set of values of parameters. Consequently, generic domain ontology containing the generic knowledge for solving problems concerning the generic process can be developed. This lead to the design of Generic Monitoring Cognitive Agent, a class of agent in which the whole knowledge corpus is generic. In other words, modeling a process within a generic space becomes configuring a generic process and adaptiveness becomes genericity, that is to say independence regarding technology. In this paper, we present an application of this approach on Sachem, a Generic Monitoring Cognitive Agent designed in order to help the operators in operating a blast furnace. Specifically, the NeuroGaz module of Sachem will be used to present the notion of a generic blast furnace. The adaptiveness of Sachem can then be noted through the low cost of the deployment of a Sachem instance on different blast furnaces and the ability of NeuroGaz in solving problem and learning from various top gas instrumentation.     

Partial equiasymptotic stability of nonlinear dynamic systems

For nonlinear nonautonomous dynamic systems, partial equiasymptotic stability of the zero equilibrium position, equiasymptotic stability of the partial equilibrium position, and partial equiasymptotic stability of the partial equilibrium position are studied by the Lyapunov functions method. Examples are given.Translated from Avtomatika i Telemekhanika, No. 2, 2005, pp. 3–16.Original Russian Text Copyright © 2005 by Alekseeva, Vorotnikov, Feofanova.     

A theory of nonmonotonic rule systems I

We introduce here the study of generalnonmonotonic rule systems. These deal with situations where a conclusion is drawn from a system of beliefsS (and seen to be inS), basedboth on some premises being inS and on some restraints not being inS. In the monotone systems of traditional logic there are no restraints, conclusions are drawn solely based on premises being inS. Nonmonotonic rule systems capture the essential syntactic, semantic, and algorithmic features of many nonmonotone systems such as default logic, negation as failure, truth maintenance, autoepistemic logic, and also important combinatorial questions from mathematics such as the marriage problem. This reveals semantics and syntax and proof procedures and algorithms for computing belief sets in many cases where none were previously available and entirely uniformly. In particular, we introduce and study deductively closed sets, extensions and weak extensions. Semantics of nonmonotonic rule systems is studied in part II of this paper and extensions to predicate classical, intuitionistic, and modal logics are left to a later paper.Work partially supported by NSF grant RII-8610671 and Kentucky EPSCoR program and ARO contract DAAL03-89-K-0124.Work partially supported by NSF grant DMS-8902797 and ARO contract DAAG629-85-C-0018.Work partially supported by NSF grant DMS-8702473.     

Modular Control and Coordination of Discrete-Event Systems

In the supervisory control of discrete-event systems based on controllable languages, a standard way to handle state explosion in large systems is by modular supervision: either horizontal (decentralized) or vertical (hierarchical). However, unless all the relevant languages are prefix-closed, a well-known potential hazard with modularity is that of conflict. In decentralized control, modular supervisors that are individually nonblocking for the plant may nevertheless produce blocking, or even deadlock, when operating on-line concurrently. Similarly, a high-level hierarchical supervisor that predicts nonblocking at its aggregated level of abstraction may inadvertently admit blocking in a low-level implementation. In two previous papers, the authors showed that nonblocking hierarchical control can be guaranteed provided high-level aggregation is sufficiently fine; the appropriate conditions were formalized in terms of control structures and observers. In this paper we apply the same technique to decentralized control, when specifications are imposed on local models of the global process; in this way we remove the restriction in some earlier work that the plant and specification (marked) languages be prefix-closed. We then solve a more general problem of coordination: namely how to determine a high level coordinator that forestalls conflict in a decentralized architecture when it potentially arises, but is otherwise minimally intrusive on low-level control action. Coordination thus combines both vertical and horizontal modularity. The example of a simple production process is provided as a practical illustration. We conclude with an appraisal of the computational effort involved.     

Playing Detective: Reconstructing Software Architecture from Available Evidence

Because a system's software architecture strongly influences its quality attributes such as modifiability, performance, and security, it is important to analyze and reason about that architecture. However, architectural documentation frequently does not exist, and when it does, it is often out of sync with the implemented system. In addition, it is rare that software development begins with a clean slate; systems are almost always constrained by existing legacy code. As a consequence, we need to be able to extract information from existing system implementations and utilize this information for architectural reasoning. This paper presents Dali, an open, lightweight workbench that aids an analyst in extracting, manipulating, and interpreting architectural information. By assisting in the reconstruction of architectures from extracted information, Dali helps an analyst redocument architectures, discover the relationship between as-implemented and as-designed architectures, analyze architectural quality attributes and plan for architectural change.     

Duality in Quantum Information Geometry

Let H be a separable Hilbert space. We consider the manifold M consisting of density operators on H such that ^p is of trace class for some p (0, 1). We say M is nearby if there exists C > 1 such that C ^–1C. We show that the space of nearby points to can be furnished with the two flat connections known as the (±)-affine structures, which are dual relative to the BKM metric. We furnish M with a norm making it into a Banach manifold.     

Feedback linearization and driftless systems

The problem of dynamic feedback linearization is recast using the notion of dynamic immersion. We investigate here a generic property which holds at every point of a dense open subset, but may fail at some points of interest, such as equilibrium points. Linearizable systems are then systems that can be immersed into linear controllable ones. This setting is used to study the linearization of driftless systems: a geometric sufficient condition in terms of Lie brackets is given; this condition is also shown to be necessary when the number of inputs equals two. Though noninvertible feedbacks are nota priori excluded, it turns out that linearizable driftless systems with two inputs can be linearized using only invertible feedbacks, and can also be put into a chained form by (invertible) static feedback. Most of the developments are done within the framework of differential forms and Pfaffian systems.This work was partially supported by INRIA, NSF Grant ECS-9203491, GR Automatique (CNRS), and DRED (Ministère de l'Éducation Nationale). Part of it was done while the first author was visiting the Center for Control Engineering and Computation, University of California at Santa Barbara.     

The impact of artificial intelligence on leisure

Two views of AI in leisure and the work-place and two views of society are discussed. There is a conceptualisation of AI systems enhancing people in their work and leisure and another of AI automata which tends to degrade and replace human activity. Researchers tend to resolve into Optimists who work within a micro-sociological view and see AI systems as inevitable and beneficent. Others are Pessimists who adopt a macro-sociological view and see AI in its automata role and deliterious social consequences. These polarised perspectives must be integrated as only enhancing AI is socially acceptable.     

Low-Rank Approximation of Integral Operators by Interpolation

A central component of the analysis of panel clustering techniques for the approximation of integral operators is the so-called -admissibility condition min {diam(),diam()} 2dist(,) that ensures that the kernel function is approximated only on those parts of the domain that are far from the singularity. Typical techniques based on a Taylor expansion of the kernel function require a subdomain to be far enough from the singularity such that the parameter has to be smaller than a given constant depending on properties of the kernel function. In this paper, we demonstrate that any is sufficient if interpolation instead of Taylor expansionisused for the kernel approximation, which paves the way for grey-box panel clustering algorithms.     

Eine bedingte Form der Integral-Ungleichung von Gronwall,leichte Verallgemeinerung eines Stabilitätssatzes von Perron und überlauffreie Skalierung von Analogrechenschaltungen

Zusammenfassung Es geht in dieser Arbeit in der Hauptsache darum, ein vorgelegtes Differentialgleichungssystem so zu skalieren, daß in der zugehörigen Analogrechnerschaltung die Spannungen an den Ausgängen der Integratoren die durch die Referenzspannung einerseits und durch das Auflösevermögen andererseits gesetzten Schranken nicht über- bzw. unterschreiten. Es werden Abschätzungssätze hergeleitet, die diese Frage im Apriori-Sinn, also ohne die Lösung des Differentialgleichungssystems zu kennen, zu lösen gestatten. Zur Abschätzung werden zunächst Normen, dannKamke-Normen verwendet. Der im Titel erwähnte Satz vonPerron ergibt sich durch spezielle Normengebung und Verzicht auf Abschätzung nach unten. Erschwert werden die Betrachtungen durch die relative Schwäche der Forderung, daß die rechte Seite des Systemsdx/dt=f(x,t) der Bedingung aus xa folgt f(x,t)v(t)x genüge (...:=Norm,a positiv reell). Dadurch scheint es bei Abschätzungen mitKamke-Normen nicht mehr möglich, von den in der Literatur über Existenzbeweise und Abschätzungssätze üblichen Methoden Gebrauch zu machen. Zur Lösung dieser Frage wird eine bedingte Form des bekannten Satzes vonGronwall (auch Satz vonBellman genannt) entwickelt.

---

A conditional version of the integral inequality of gronwall, a slight generalization of a stability theorem of perron, and overflow-free scaling of analogue computer set-ups

Summary The main subject of this paper is the scaling of a given set of differential equations in such a way that the output voltages of the integrators of the associated analogue computer set-up do not exceed certain upper and lower bounds imposed by the reference voltage and the limited power of resolution of the elements of the analogue computer. The paper gives a priori bounds on the solution of the differential set. Some of these bounds work with norms, others withKamke-norms.Perron's stability theorem mentioned in the title of this paper results by inserting special norms and neglecting lower bounds. A difficulty arises by the relative weakness of the condition xa implies f(x,t)v(t)x on the right hand side of the setdx/dt=f(x,t), where ... is any norm anda is a positive real constant. As a consequence of this, it seems no longer possible to use the usual techniques known from the literature on existence theorems and bounds for the solution of differential equations. To cope with this situation, a conditional version of the well-known theorem ofGronwall (also known by the name of Lemma ofBellman) will be derived.

---

Diese Arbeit ist Teil einer am Institut für Angewandte Mathematik der Technischen Hochschule München unter Anleitung von Herrn o. Prof. Dr. rer. nat. habil.J. Heinhold angefertigten Dissertation.     

Universal Adaptive Stabilization of Nonlinear Systems

Mårtensson's result The order of a stabilizing regulator issufficient a priori information for adaptive stabilization is proved to bevalid also for a class of nonlinear, time–varying systems and thefeedback strategy is simplified.     

On the nature of rules and conversation

The use of findings from conversation analysis in the design of human-computer interfaces and especially in the design of computer-human speech dialogues is a matter of considerable controversy. For example, in Going up a Blind Alley (Button, 1990) and On Simulacrums of Conversation (Button and Sharrock, 1995), Button argues that conversation analysis is of only limited use in the computational modelling of interaction. He suggests that computers will never be able to converse with humans because of the fundamentally different ways in which humans and computers use rules in the production of language.We show in this paper that these arguments are neither necessary nor sufficient to rule out the possibility of computers which can be said to converse. They depend on a view about the nature of rules which is based on a fundamental misunderstanding of the scope of computation. The way in which mathematical systems such as Context Free Grammars use rules is very different from the use of the rules in principle-based approaches to language or the micro-rules of neural networks. If there is a problem with conversing computers, it lies more with the true nature of the interaction that is taking place and with considerations about the nature of cognition than with the construction and use of rules.