Wannes Meert

Lifted probabilistic inference by first-order knowledge compilation

Probabilistic logical languages provide powerful formalisms for knowledge representation and learning. Yet performing inference in these languages is extremely costly, especially if it is done at the propositional level. Lifted inference algorithms, which avoid repeated computation by treating indistinguishable groups of objects as one, help mitigate this cost. Seeking inspiration from logical inference, where lifted inference (e.g., resolution) is commonly performed, we develop a model theoretic approach to probabilistic lifted inference. Our algorithm compiles a first-order probabilistic theory into a first-order deterministic decomposable negation normal form (d-DNNF) circuit. Compilation offers the advantage that inference is polynomial in the size of the circuit. Furthermore, by borrowing techniques from the knowledge compilation literature our algorithm effectively exploits the logical structure (e.g., context-specific independencies) within the first-order model, which allows more computation to be done at the lifted level. An empirical comparison demonstrates the utility of the proposed approach.

SessionShield: lightweight protection against session hijacking

The class of Cross-site Scripting (XSS) vulnerabilities is the most prevalent security problem in the field of Web applications. One of the main attack vectors used in connection with XSS is session hijacking via session identifier theft. While session hijacking is a client-side attack, the actual vulnerability resides on the server-side and, thus, has to be handled by the websites operator. In consequence, if the operator fails to address XSS, the applications users are defenseless against session hijacking attacks. In this paper we present SessionShield, a lightweight client-side protection mechanism against session hijacking that allows users to protect themselves even if a vulnerable websites operator neglects to mitigate existing XSS problems. SessionShield is based on the observation that session identifier values are not used by legitimate clientside scripts and, thus, need not to be available to the scripting languages running in the browser. Our system requires no training period and imposes negligible overhead to the browser, therefore, making it ideal for desktop and mobile systems.

Range and coexistence analysis of long range unlicensed communication

A broad range of emerging applications require very low power, very long range yet low throughput communication. Different standards are being proposed to meet these novel requirements. In this paper, the technical differences between a wideband spread spectrum (LoRa-like) and an ultra narrowband (Sigfox-like) network will be explained and evaluated. On the physical layer, simulation results show that an ultra narrowband network has a larger coverage, while wideband spread spectrum networks are less sensitive to interference. When considering the contention between nodes and interference between different networks, simulations show that adaptation of frequency and modulation is imperative for efficiently dealing with varying contention and interference in long range unlicensed networks. Depending on network load, size and distance, a device in a wideband network can send 6 times more packets to the base station when there is active rate and frequency management and an intra-technology control plane.

LS-SVM based spectral clustering and regression for predicting maintenance of industrial machines

Abstract Accurate prediction of forthcoming faults in modern industrial machines plays a key role in reducing production arrest, increasing the safety of plant operations, and optimizing manufacturing costs. The most effective condition monitoring techniques are based on the analysis of historical process data. In this paper we show how Least Squares Support Vector Machines (LS-SVMs) can be used effectively for early fault detection in an online fashion. Although LS-SVMs are existing artificial intelligence methods, in this paper the novelty is represented by their successful application to a complex industrial use case, where other approaches are commonly used in practice. In particular, in the first part we present an unsupervised approach that uses Kernel Spectral Clustering (KSC) on the sensor data coming from a vertical form seal and fill (VFFS) machine, in order to distinguish between normal operating condition and abnormal situations. Basically, we describe how KSC is able to detect in advance the need of maintenance actions in the analysed machine, due the degradation of the sealing jaws. In the second part we illustrate a nonlinear auto-regressive (NAR) model, thus a supervised learning technique, in the LS-SVM framework. We show that we succeed in modelling appropriately the degradation process affecting the machine, and we are capable to accurately predict the evolution of dirt accumulation in the sealing jaws.

CP-logic theory inference with contextual variable elimination and comparison to BDD based inference methods

There is a growing interest in languages that combine probabilistic models with logic to represent complex domains involving uncertainty. Causal probabilistic logic (CP-logic), which has been designed to model causal processes, is such a probabilistic logic language. This paper investigates inference algorithms for CP-logic; these are crucial for developing learning algorithms. It proposes a new CP-logic inference method based on contextual variable elimination and compares this method to variable elimination and to methods based on binary decision diagrams.

Chr(prism)-based probabilistic logic learning

PRISM is an extension of Prolog with probabilistic predicates and built-in support for expectation-maximization learning. Constraint Handling Rules (CHR) is a high-level programming language based on multi-headed multiset rewrite rules. In this paper, we introduce a new probabilistic logic formalism, called CHRiSM , based on a combination of CHR and PRISM. It can be used for high-level rapid prototyping of complex statistical models by means of “chance rules”. The underlying PRISM system can then be used for several probabilistic inference tasks, including probability computation and parameter learning. We define the CHRiSM language in terms of syntax and operational semantics, and illustrate it with examples. We define the notion of ambiguous programs and define a distribution semantics for unambiguous programs. Next, we describe an implementation of CHRiSM , based on CHR(PRISM). We discuss the relation between CHRiSM and other probabilistic logic programming languages, in particular PCHR. Finally, we identify potential application domains.

Bitsquatting: exploiting bit-flips for fun, or profit?

Over the last fifteen years, several types of attacks against domain names and the companies relying on them have been observed. The well-known cybersquatting of domain names gave way to typosquatting, the abuse of a users mistakes when typing a URL in her browsers address bar. Recently, a new attack against domain names surfaced, namely bitsquatting. In bitsquatting, an attacker leverages random bit-errors occurring in the memory of commodity computers and smartphones, to redirect Internet traffic to attacker-controlled domains. In this paper, we report on a large-scale experiment, measuring the adoption of bitsquatting by the domain-squatting community through the tracking of registrations of bitsquatting domains targeting popular web sites over a 9-month period. We show how new bitsquatting domains are registered daily and how attackers are trying to monetize their domains through the use of ads, abuse of affiliate programs and even malware installations. Lastly, given the discovered prevalence of bitsquatting, we review possible defense measures that companies, software developers and Internet Service Providers can use to protect against it.

24.2 Context-aware hierarchical information-sensing in a 6μW 90nm CMOS voice activity detector

The rise of always-listening sensors integrated in energy-scarce devices such as watches and remote-controls increases the need for intelligent scalable interfaces. Contemporary sensor interfaces digitize raw sensor data to extract information with energy-intensive computations, such as FFT, which is inefficient if the end goal is to only extract selective information for classification tasks, e.g. voice activity detection (VAD). Previous work shows energy gains from early data reduction through analog feature extraction [1] or embedded classification hardware [2]. However, the potential energy savings of these devices is limited as they cannot adapt to changes in the sensed information content or sensing context, such as the amount/type of acoustic background noise. In the processor design community, such adaptivity to varying operating conditions is actively researched through the concept of hierarchical computing [3]. This work integrates the concept of hierarchical operation with adaptive early data extraction and classification, towards a power- and context-aware information-extraction sensor interface. This paper specifically reports on a μW 90nm CMOS VAD, that dynamically adapts sensing resources to signal information content and context, thus only spending energy on relevant information extraction. An order of magnitude in power savings is achieved by exploiting hierarchical sensing, run-time activated/scalable analog feature extraction and tightly-integrated context-aware mixed-signal machine learning inference, enabling novel applications in area of acoustic sensing [1,4].

A 90 nm CMOS,

As the number of people above 65 continuously grows the demand for appropriate support to allow this group of people to live independently increases as well. Consequently, a lot of research effort is focused on the development of new technologies that can provide this support. In contrast, only a limited number of these new developments are successfully launched on the healthcare market. In order to facilitate this penetration of the healthcare market, an intense collaboration strategy between healthcare workers, older adults, informal caregivers and engineers is proposed in this paper.

6\ {\upmu {\text{W}}}

Logic programs with annotated disjunctions, or LPADs, are an elegant knowledge representation formalism that can be used to combine first order logical and probabilistic inference. While LPADs can be written manually, one can also consider the question of how to learn them from data. Methods for learning restricted classes of LPADs have been proposed before, but the problem of learning any kind of LPADs was still open. In this paper, we describe a reduction of non-recursive LPADs with a finite Herbrand universe to Bayesian networks. This reduction makes it possible to learn such LPADs using standard learning techniques for Bayesian networks. Thus the class of learnable LPADs is extended.