Robust detection of outliers for projection-based face recognition methods

In this paper, the impact of outliers on the performance of high-dimensional data analysis methods is studied in the context of face recognition. Most of the existing face recognition methods are based on PCA-like methods: faces are projected into a lower dimensional space in which similarity between faces is supposed to be more easily evaluated. These methods are, however, very sensitive to the quality of the face images used in the training and in the recognition phases. Their performance significantly drops when face images are not well centered or taken under variable illumination conditions. In this paper, we study this phenomenon for two face recognition methods, namely PCA and LDA2D, and we propose a filtering process that allows the automatic selection of noisy face images which are responsible for the performance degradation. This process uses two techniques. The first one is based on the recently proposed robust high-dimensional data analysis method called RobPCA. It is specific to the case of recognition from video sequences. The second technique is based on a novel and effective face classification technique. It allows isolating still face images that are not very precisely cropped, not well-centered or in a non-frontal pose. Experiments show that this filtering process significantly improves recognition rates by 10 to 30%.

Kinetics of Multi‐Step Redox Processes by Time‐Resolved In Situ X‐ray Diffraction

The kinetics of redox reactions of iron oxide in oxygen carrier 50Fe₂O₃/MgAl₂O₄ are examined using different time‐resolved techniques. Reduction kinetics are studied by H₂ temperature‐programmed reduction (H₂‐TPR) monitored by time‐resolved in situ XRD. In contrast to conventional TPR, in situ XRD distinguishes the three‐stage reduction of Fe₂O₃ → Fe₃O₄ → FeO → Fe. It also shows that the oxidation of Fe → Fe₃O₄ by CO₂ has no intermediate crystalline phases, explaining why its kinetics can easily be investigated by conventional CO₂ temperature‐programmed oxidation (CO₂‐TPO). A shrinking core model which takes into account solid state diffusion allows describing the experimental data.     

Exact Sampling of a Linear Interval Predictor

Recent advances in the design of interval observers have made it possible to ensure the non‐divergence of the computed state bounds from the stability of LTI systems under bounded inputs, with no need for additional monotony assumptions. Time‐varying changes of coordinates can be used to that purpose. Most of the related works result in either continuous‐time or discrete‐time interval dynamics. This paper proposes a constructive algorithm to compute the exact sampled response of a linear interval predictor under bounded inputs, gives a stability equivalence result and discusses the design of interval observers. The exact sampling requires held input bounds but the uncertain input itself needs not to be held. A numerical example exhibiting an oscillatory behavior illustrates the main results.     

Using duration models to reduce fragmentation in audio segmentation

We investigate explicit segment duration models in addressing the problem of fragmentation in musical audio segmentation. The resulting probabilistic models are optimised using Markov Chain Monte Carlo methods; in particular, we introduce a modification to Wolff’s algorithm to make it applicable to a segment classification model with an arbitrary duration prior. We apply this to a collection of pop songs, and show experimentally that the generated segmentations suffer much less from fragmentation than those produced by segmentation algorithms based on clustering, and are closer to an expert listener’s annotations, as evaluated by two different performance measures. Editor: Gerhard Widmer     

Monte Carlo simulation of 2D TASER

A possibility to develop the so called TASER (Terahertz-Amplification-by-the-Stimulated-Emission-of- Radiation) by using two-dimensional (2D) electron transport in quantum well (QW) structures is investigated by Monte Carlo simulation of the optical-phonon-emission assisted transit-time resonance (OPTTR) of 2D electrons in momentum space under the low lattice temperature. A considerable extension of the frequency region for THz radiation generation (upto 5 times) when going from 3D- to 2D-case is predicted.     

Siamese multi-layer perceptrons for dimensionality reduction and face identification

This paper presents a framework using siamese Multi-layer Perceptrons (MLP) for supervised dimensionality reduction and face identification. Compared with the classical MLP that trains on fully labeled data, the siamese MLP learns on side information only, i.e., how similar of data examples are to each other. In this study, we compare it with the classical MLP on the problem of face identification. Experimental results on the Extended Yale B database demonstrate that the siamese MLP training with side information achieves comparable classification performance with the classical MLP training on fully labeled data. Besides, while the classical MLP fixes the dimension of the output space, the siamese MLP allows flexible output dimension, hence we also apply the siamese MLP for visualization of the dimensionality reduction to the 2-d and 3-d spaces.     

Low response time context awareness through extensible parameter adaptation with ORCA

Ubiquitous computing applications or widespread robots interactions execute in unforeseen environments and need to adapt to changeful available services, user needs, and variations of the environment. Context-awareness ability addresses such a need, enabling, through adaptation rules, applications to react to the perceived dynamic variations. Responses to adaptation have to be quick enough to maximize the time during which the application is coherent with its environment. Adaptation rules, associating variations of the environment to application reactions, are usually established at design time. However, in unforeseen and partially anticipated environments, we claim that adaptation rules have to be dynamically extensible to match previously unexpected variations. Our approach enables rule composition and ensures a deterministic result. We propose to use parameter adaptation to quickly respond to environmental variations and dynamic compositional adaptation to provide extensibility to the parameter adaptation. To foster even lower response times, we internalize context-awareness processing and decision into the application.     

Effects of mechanical interactions on the hydrostatic stress in randomly distributed rubber particles in an amorphous polymer matrix

The equivalent inclusion method (EIM) assuming linear elasticity is used to calculate the mechanical interactions between spherical rubber particles in an amorphous matrix, as in a rubber toughened polymer. The influences of the various calculation parameters are examined and it is shown that the method can provide reliable results with regard to the level of hydrostatic stress in the particles. Damage of the material is simulated by replacing the most stressed particles by voids. Numerical simulations for several hundreds of interacting particles give information on the kinetics and spatial organisation of the damage. It appears that, as the volume fraction of particles increases from 10 to 20%, the spatial configuration of the damage evolves from a localised to a diffuse mode. These results are discussed in relation to the efficiency of rubber toughening.     

Physicochemical Properties of Palm Oil-Based Puff Pastry Model Margarines Related to Their Baking Performance in Long-Term Storage

Developing trans-free alternative fat solutions suitable for specific applications remains a challenge in edible fats and other domains. This is particularly true for palm oil-based puff pastry margarines, which suffer from post crystallization problems, leading to dramatic loss of functionality. This research is aimed at investigating the influence of triacylglycerol (TAG) compositions of palm oil-based puff pastry margarines on the physical properties of the fat crystal network, which determine the functionality of such products. Three model puff pastry margarines are produced at pilot scale under the same crystallization conditions. They share the same fatty acid composition and close solid fat content (SFC) profiles, whereas the proportions of major TAG (tripalmitoylglycerol (PPP), 1,3-di-palmitoyl-2-oleoylglycerol (POP), 1,2-di-palmitoyl-3-oleoylglycerol (PPO), 1,2-dioleoyl-3-palmitoylglycerol (POO)) are different. Polymorphism, melting profile, hardness, microscopic structures, and baking performance (puffing effect) of the model fats are examined during a period of 6 months. The following results are obtained: 1) The TAG composition significantly influences the post crystallization processes occurring in palm oil-based margarines. 2) High amounts of POP show negative influences. 3) The proportions of POP, PPO, and PPP should be carefully balanced to prevent detrimental crystal network rearrangements, leading to textural modifications (hardness increase) and significant reduction in baking performance. Practical Applications : The results presented in this work could be helpful for edible fat products developers, especially for roll-in fat applications. This research provides an overview of the relevant properties to study for the assessment of puff pastry margarine functionality. It also highlights the importance of ensuring long-term stability of palm oil-based fat products. Finally, it emphasizes that certain combinations of fat materials should be avoided to maintain the quality of palm oil-based puff pastry margarines.     

Twin mirrors for laser interferometric gravitational-wave detectors

Gravitational-wave detectors such as Virgo and the laser interferometric gravitational-wave observatory (LIGO) use a long-baseline Michelson interferometer with Fabry-Perot cavities in the arms to search for gravitational waves. The symmetry between the two Fabry-Perot cavities is crucial to reduce the interferometer's sensitivity to the laser amplitude and frequency noise. To this purpose, the transmittance of the mirrors in both cavities should be as close as possible. This paper describes the realization and the characterization of the first twin large low-loss mirrors with transmissions differing by less than 0.01%.