Search

Sensors and systems often exhibit different dynamics in response to various stimuli. The problem of determining which of the dynamics has generated a given observation sequence is referred to as a multiclass discrimination problem. We propose a solution to this problem as a minimum distance classifier that assigns the observation sequence to the dynamics to which it is closest in terms of a similarity metric. One of the main contributions is to compute this metric in a high-dimensional feature space using the Kronecker product so that the projected data of each class lie on distinct hyperplanes. These are trained using singular value decomposition or quadratic programing in the spirit of twin support vector machines (SVMs). The other main contribution is to show that the minimum distance classifier exists for state-space models if and only if the observability matrices are all different even when the models are not observable in a control-theoretic sense. We validated our method on numerical examples of dynamical systems and on gas identification using metal-oxide sensors. Results confirm the theoretical condition for the existence of the classifier and demonstrate the efficiency of the learning process with classification accuracy matching or exceeding state-of-the art performance. In particular, learning the classifier directly from training data is more robust and more generally applicable than determining it from classical system identification methods. The minimum distance classifier is thus relevant to the classification of dynamical sensing systems.

Controlling over-constrained cable-driven parallel robots (CDPRs) is a challenging task due to the complex dynamics of the system. Classical controllers require force distribution algorithms that involve an optimization problem, which is time consuming. In this paper, we propose an AI-based approach that learns a controller from simulated trajectories. A dynamic model of the CDPR is first validated experimentally on a real robot. Then, the controller is trained on the CDPR simulator with randomly generated trajectories using a deep deterministic policy gradient (DDPG). Finally, the trained controller is tested on different trajectories. Validation results show that the proposed approach is able to track unknown trajectories with a good accuracy.

It is believed that humans and animals like rodents and bats navigate in a familiar environment using a cognitive map. Yet, how maps are previously learned when exploring a novel environment and then used to plan routes to specific goals remains unclear. We propose here a biologically inspired method, called Multiwave, with two main ingredients: (i) a symmetric spike-timing dependent plasticity rule that compels the connectivity to map the environment, and (ii) periodic wavefronts propagating through the network to direct the agent towards the goal and increase robustness to noise. The modeling involves a detailed neural network for biological plausibility and a simplified equivalent network, better suited for robotic implementation. Using both simulations and robotic experiments, we show the efficiency of Multiwavefor path planning. Multiwaveis relevant not only to neuroscience, as the detailed model is biologically grounded, but also to robotic applications, as the simplified model is suitable for analog implementation on neuromorphic hardware.

Two basic tasks must be performed by an olfactory robot tracking a specific odor source: navigate in a turbulent odor plume and recognize an odor regardless of its concentration. For these two tasks, we propose simple biologically inspired strategies, well suited for building dedicated circuits and for on-board implementation on real robots. The odor recognition system is based on a spiking neural network using a synchronization coding scheme. The robot navigation system is based on the use of bilateral comparison between two spatially separated gas sensors arrays at either side of the robot. We propose binary or analog navigation laws depending on the nature of the available sensory information extracted from the plume structure (isolated odor patches or smoother concentration field).

We established a set of 30 microsatellites of Bovidae origin for use in a biodiversity study in Swiss and Creole goats. Additional microsatellites located within or next to ”candidate” genes of interest, such as cytokine genes (IL4, INF-gamma) and MHC class II genes (DRB, DYA) were tested in the caprine species in order to detect possible associations with two infectious caprine diseases. Microsatellite analysis was undertaken using automated sequencers (ABI373 & 3100). In the first study, a total of 82 unrelated Creole goats, 37 resistant and 45 susceptible to Heartwater disease (Cowdriosis) were analysed. In this study, the two microsatellite loci DRBP1 (MHCII) and BOBT24 (IL4) were positively associated with disease susceptibility, demonstrating a corrected $P$-value of 0.002 and 0.005, respectively. In a second investigation, we tested 36 goats, naturally infected with the nematode parasite Trichostrongylus colubriformis. These animals were divided into a ”low” and ”high” excreting group on the basis of two independently recorded fecal egg counts. For this nematode resistance study, we detected a significant association of one of the alleles of the microsatellite locus SPS113 with ”low” excretion (resistance). The MHC class II locus DYA (P19), was weakly associated with susceptibility in both diseases ($P_{\rm c} = 0.05$). In future experiments, we will extend the sample size in order to verify the described associations.

Flow cytometry (FCM) provides an alternative to radioactive methods for the analysis of T-cell responses. However, a comparative study of common FCM assays in an outbred ruminant model is lacking, which motivated this work. Goats immunized with the obligate intracellular bacterium Cowdria ruminantium, inactivated and emulsified in oil-based adjuvants, were used as a model to study T-cell recall responses in vitro. FCM-based methods to measure Cowdria-induced lymphoblastogenesis, DNA synthesis, and interleukin-2 receptor (IL-2R) expression by T-cell subsets were compared. IL-2R expression was the most sensitive and reliable method provided that the number of molecules per cell was analyzed and not simply the percentage of positive cells of a given phenotype. Despite high background due to adjuvant and low proliferation, this method could detect antigen-specific activation of immune CD4⁺ and CD8⁺ T cells. FCM-based measurement of lymphoblastogenesis and DNA synthesis are not the most appropriate methods to analyze T-lymphocyte activation during vaccination of outbred animals. On several occasions, analysis of IL-2R expression was the only assay capable of discriminating between vaccinated and naive animals in this model. Cytometry 49:49–55, 2002. © 2002 Wiley-Liss, Inc.

Cowdria ruminantiumis the etiologic agent of heartwater, a tick-transmitted foreign animal disease with considerable potential for entrance into the USA. A competitive enzyme-linked immunosorbent assay (cELISA) was developed to detect serologic responses to C. ruminantiuminfection. The cELISA utilized a recombinant form of the C. ruminantiummajor antigenic protein (MAP-1) as the antigen and an anti-MAP-1 monoclonal antibody as the competing indicator reagent. Experimental antisera to C. ruminantiumand a wide variety of related ehrlichial organisms were used to evaluate cELISA reactivity. Only sera against C. ruminantium, Ehrlichia canis, E. chaffeensis, and a recently discovered cervine ehrlichia-like organism reacted positively in the cELISA. Specificity of the cELISA was 299.5% in a survey of 1,774 southeastern US and Puerto Rican slaughter cattle sera but was only 85% in a group of 79 hunter-killed white-tailed deer (Odocoileus virginianus) from the southeastern USA. Reference true-positive and cELISA false-positive sera were further analyzed by end point titrations using the cELISA and by indirect fluorescent antibody (IFA) tests for reactivity with C. ruminantium, E. canis, and E. chaffeensisantigens. True heartwater-positive sera were significantly more reactive using the cELISA and C. ruminantiumIFA procedures (P< 0.05), whereas false-positive sera were significantly more reactive with the antigens used in the E. chaffeensisIFA procedure (P< 0.05). A group of sera from 210 field-origin ruminants residing on known or potentially heartwater-endemic Caribbean islands revealed a substantial (12.4%) prevalence of cELISA-positive specimens. The cELISA is a relatively specific serodiagnostic test for heartwater in cattle and could be used to monitor for possible introduction of the disease into the USA. The cELISA may also be an excellent tool for monitoring the success of an ongoing Caribbean Amblyommatick eradication program designed to eliminate the biological vector responsible for the perpetuation and spread of this dangerous foreign animal disease.

A novel unsupervised learning rule, called Boundary Adaptation Rule (BAR), is introduced for scalar quantization. It is shown that the rule maximizes information-theoretic entropy and thus yields equiprobable quantizations of univariate probability density functions. It is shown by simulations that BAR outperforms other unsupervised competitive learning rules in generating equiprobable quantizations. It is also shown that our rule can do better or worse than the Lloyd I algorithm in minimizing average mean square error, depending on the input distribution. Finally, an application to adaptive non-uniform analog to digital (A/D) conversion is considered.

Sixteen primers were successfully used in a RAPD assay to generate reproducible fingerprints for six isolates of Cowdria ruminantium, a tick‐transmitted rickettsia of ruminants. Distinction between stocks was possible by using one or at most two primers. Two stocks were very similar although originating from widely distant geographical regions. A genetic distance tree was constructed by analysing 108 fragments in pairwise comparison between stocks. Three amplification fragments probed with C. ruminantiumgenomic DNA determined a restriction fragment length polymorphism which allowed the distinction between stocks except for the two stocks that had similar RAPD patterns. The potential of RAPD to determine the extent of genetic diversity of C. ruminantiumand to develop probes or PCR primers for diagnostic purposes is discussed.

We established a set of 30 microsatellites of Bovidae origin for use in a biodiversity study in Swiss and Creole goats. Additional microsatellites located within or next to "candidate" genes of interest, such as cytokine genes (IL4, INF-gamma) and MHC class II genes (DRB, DYA) were tested in the caprine species in order to detect possible associations with two infectious caprine diseases. Microsatellite analysis was undertaken using automated sequencers (ABI373 & 3100). In the first study, a total of 82 unrelated Creole goats, 37 resistant and 45 susceptible to Heartwater disease (Cowdriosis) were analysed. In this study, the two microsatellite loci DRBP1 (MHCII) and BOBT24 (IL4) were positively associated with disease susceptibility, demonstrating a corrected P-value of 0.002 and 0.005, respectively. In a second investigation, we tested 36 goats, naturally infected with the nematode parasite Trichostrongylus colubriformis. These animals were divided into a "low" and "high" excreting group on the basis of two independently recorded fecal egg counts. For this nematode resistance study, we detected a significant association of one of the alleles of the microsatellite locus SPS113 with "low" excretion (resistance). The MHC class II locus DYA (P19), was weakly associated with susceptibility in both diseases (Pc= 0.05). In future experiments, we will extend the sample size in order to verify the described associations.

Sixteen primers were successfully used in a RAPD assay to generate reproducible fingerprints for six isolates of Cowdria ruminantium, a tick-transmitted rickettsia of ruminants. Distinction between stocks was possible by using one or at most two primers. Two stocks were very similar although originating from widely distant geographical regions. A genetic distance tree was constructed by analysing 108 fragments in pairwise comparison between stocks. Three amplification fragments probed with C. ruminantium genomic DNA determined a restriction fragment length polymorphism which allowed the distinction between stocks except for the two stocks that had similar RAPD patterns. The potential of RAPD to determine the extent of genetic diversity of C. ruminantium and to develop probes or PCR primers for diagnostic purposes is discussed.