AUTOMATION OF GENERALIZED ADDITIVE NEURAL NETWORKS FOR PREDICTIVE DATA MINING

For a new technology to make the step from experimental technology to mainstream technology, tools need to be created to facilitate the use of the developed technology in the envisaged application area. Generalized additive neural networks provide an attractive framework that shows promise in the field of predictive data mining. However, the construction of such networks is very time consuming and subjective, because it depends on the user to interpret partial residual plots and to make changes in the neural network architecture. For this technology to be accepted as a serious modeling option in the field of predictive data mining the construction process needs to be automated and the benefits of using the technique must be clearly illuminated. This article shows how intelligent search may be used to replace subjective human judgment with objective criteria and make generalized additive neural networks an attractive option for the modeler.     

Application of a bacteriological on-farm test to reduce antimicrobial usage in dairy cows with purulent vaginal discharge

The objective of this study was to assess the effect of a selective antibiotic treatment strategy based on a quick bacteriological on-farm test (Petrifilm, 3M Corp., St. Paul, MN) compared with the conventional antibiotic treatment of all cows having clinical endometritis (CE) defined by the presence of purulent vaginal discharge on both clinical cure rate and reproductive performance. The study was simultaneously conducted with dairy cows reared under a highly supplemented rotational grazing system in Argentina and in a freestall system in Slovakia. Cows having an abnormal vaginal discharge (VD, indicative of clinical endometritis) on 21 to 35 d in milk (DIM) were randomly allocated to 1 of 2 study groups: selective treatment (ST) or conventional treatment (CT). All cows in the CT group (n = 174) received a single intrauterine administration of 500 mg of cephapirin. In the ST group (n = 178), treatment decision was made according to the results of the bacteriological on-farm test. For this test, we collected intrauterine samples with the cytobrush technique and stroke the brushes onto 2 different Petrifilm plates, one for aerobic count and another for Enterobacteriaceae count, incubated the plates, and counted the number of colonies after 24 h. Positive cows (≥5 colonies in one or both plates) received a single intrauterine treatment with 500 mg of cephapirin, whereas negative cows (<5 colonies) remained untreated. Clinical cure rate was assessed by direct vaginal inspection at 14 d after treatment (VD-0). The odds for conception at first artificial insemination, artificial insemination by 80 DIM, pregnancy by 100 DIM, and for nonpregnancy by 200 DIM were estimated with mixed logistic regression models. The hazard of conception was also assessed with proportional hazard regression model. The selective antibiotic treatment strategy based on the outcome of Petrifilm test reduced the number of required treatments (57%) and maintained similar efficacy in terms of clinical cure and reproductive performance as the conventional antibiotic treatment of all endometritic cows.     

A statistical framework for the analysis of long image time series

Coarse spatial resolution satellites are capable of observing large swaths of the planetary surface in each overpass resulting in image time series with high temporal resolution. Many change‐detection strategies commonly used in remote sensing studies were developed in an era of image scarcity and thus focus on comparing just a few scenes. However, change analysis methods applicable to images with sparse temporal sampling are not necessarily efficient and effective when applied to long image time series. We present a statistical framework that gathers together: (1) robust methods for multiple comparisons; (2) seasonally corrected Mann–Kendall trend tests; (3) a testing sequence for quadratic models of land surface phenology. This framework can be applied to long image time series to partition sources of variation and to assess the significance of detected changes. Using a standard image time series, the Pathfinder AVHRR Land (PAL) NDVI data, we apply the framework to address the question of whether the institutional changes accompanying the collapse of the Soviet Union resulted in significant changes in land surface phenologies across the ecoregions of Kazakhstan.     

A functional approach to modelling of manufacturing processes

A functional model of batch processes is formulated. The model was derived from considerations of multi-stage decision processes when some degree of optimization is required. Consistent with system theory and dynamic programming techniques, the model was derived by considering the overall system output and working backwards in time, or order of command priority. To facilitate this mode of thinking the process was divided into two basic functions, enabling the final stage to be treated in a linear, autonomous way. Application of the definition of optimality indicated that the final stage of the model is of paramount importance and produces a logical starting point for mathematical and managerial decision making policies.     

Fractional-order filters for active damping in a lithographic tool

In a lithographic tool used in semiconductor manufacturing, vibrations in the projection lens reduce image contrast and can limit the system's resolution. Active damping by direct velocity feedback can reduce this effect, but requires sufficient amplitude roll-off for higher frequencies in the open-loop transfer function. Ongoing mass decoupling, however, tends to increase the transfer function magnitude. Stability is endangered by phase lag induced by control loop components and non-collocation of the actuator or sensor. Conventional lowpass filtering is not possible because of the minimum induced 90° phase shift. This paper introduces the usage of fractional-order filters with an order smaller than one in the feedback controller. The filter's reduced phase lag allows amplitude reduction for higher frequencies without stability risk. Implementation of 1/2-order and 2/3-order filters, and their effect in an active damping system for a lithographic tool's projection lens, are shown.     

Modeling and performance analysis of an absorption chiller with a microchannel membrane‐based absorber using LiBr‐H2O,LiCl‐H2O,and LiNO3‐NH3

In order to develop compact absorption refrigeration cycles driven by low heat sources, the simulated performance of a microchannel absorber of 5‐cm length and 9.5 cm³ in volume provided with a porous membrane is presented for 3 different solution‐refrigerant pairs: LiBr‐H₂O, LiCl‐H₂O, and LiNO₃‐NH₃. The high absorption rates calculated for the 3 solutions lead to large cooling effect to absorber volume ratios: 625 kW/m³ for the LiNO₃‐NH₃, 552 kW/m³ for the LiBr‐H₂O, and 318 kW/m³ for the LiCl‐H₂O solutions given the studied geometry. The performance of a complete absorption system is also analyzed varying the solution concentration, condensation temperature, and desorption temperature. The LiNO₃‐NH₃ and the LiBr‐H₂O solutions provide the largest cooling effects. The LiNO₃‐NH₃ can work at a lower temperature of the heating source, in comparison with the one needed in a LiBr‐H₂O system. The lowest cooling effect and coefficient of performance are found for the LiCl‐H₂O solution, but this mixture allows the use of lower temperature heating sources (below 70°C). These results can be used for the selection of the most suitable solution for a given cooling duty, depending on the available heat source and condensation temperature.     

Energy harvesting with pairs of variable capacitors without control circuits

This work describes several electronic electrostatic generators that can be built using two pairs of complementary variable capacitors, in a way that dispenses control circuits for their operation. In all cases, a basic unstable generator based on an electrostatic charge multiplier is used to bias variable capacitors, and the current generated by these capacitors is then rectified and sent to the load. Experimental versions of the proposed generators were built using variable capacitors made with 3D-printing techniques, moved in a back-and-forth way by a small motor. As these generators operate at high impedance level, always with small currents, they are insensitive to the resistivity of the conductive plastic used to make the variable capacitors. Several new structures are presented, and their properties are evaluated theoretically and experimentally.     

Universality in ant behaviour

Prediction for social systems is a major challenge. Universality at the social level has inspired a unified theory for urban living but individual variation makes predicting relationships within societies difficult. Here, we show that in ant societies individual average speed is higher when event duration is longer. Expressed as a single scaling function, this relationship is universal because for any event duration an ant, on average, moves at the corresponding average speed except for a short acceleration and deceleration at the beginning and end. This establishes cause and effect within a social system and may inform engineering and control of artificial ones.