Scaling the Indian Buffet Process via Submodular Maximization

Inference for latent feature models is inherently difficult as the inference space grows exponentially with the size of the input data and number of latent features. In this work, we use Kurihara & Welling (2008)'s maximization-expectation framework to perform approximate MAP inference for linear-Gaussian latent feature models with an Indian Buffet Process (IBP) prior. This formulation yields a submodular function of the features that corresponds to a lower bound on the model evidence. By adding a constant to this function, we obtain a nonnegative submodular function that can be maximized via a greedy algorithm that obtains at least a one-third approximation to the optimal solution. Our inference method scales linearly with the size of the input data, and we show the efficacy of our method on the largest datasets currently analyzed using an IBP model.

Influence of selected Indian immunostimulant herbs against white spot syndrome virus (WSSV) infection in black tiger shrimp, Penaeus monodon with reference to haematological, biochemical and immunological changes

Immunostimulants are the substances, which enhance the non-specific defence mechanism and provide resistance against the invading pathogenic micro-organism. In order to increase the immunity of shrimps against the WSSV, the methanolic extracts of five different herbal medicinal plants like Cyanodon dactylon, Aegle marmelos, Tinospora cordifolia, Picrorhiza kurooa and Eclipta alba were selected and mixed thoroughly in equal proportion. The mixed extract was supplemented with various concentrations viz. 100 (A), 200 (B), 400 (C), and 800 (D) mg kg(-1) through artificial diets individually. The prepared diets (A-D) were fed individually to WSSV free healthy shrimp Penaeus monodon with an average weight of 8.0 +/- 0.5 g for 25 days. Control diet (E), devoid of herbal extract was also fed to shrimps simultaneously. After 25 days of feeding experiment, the shrimps were challenged with WSSV, which were isolated and propagated from the infected crustaceans. The shrimps succumbed to death within 7 days when fed on no herbal immunostimulant diet (E). Among the different concentrations of herbal immunostimulant supplemented diets, the shrimps fed on diet D (800 mg kg(-1)) significantly (P < 0.0001) had more survival (74%) and reduction in the viral load. Also the better performance of haematological, biochemical and immunological parameters was found in the immunostimulant incorporated diets fed shrimps. The present work revealed that the application of herbal immunostimulants will be effective against shrimp viral pathogenesis and they can be recommended for shrimp culture. (c) 2006 Published by Elsevier Ltd.

Effect of low-frequency Rossby wave on thermal structure of the upper southwestern tropical Indian Ocean

We investigate the influence of low-frequency Rossby waves on the thermal structure of the upper southwestern tropical Indian Ocean (SWTIO) using Argo profiles, satellite altimetric data, sea surface temperature, wind field data and the theory of linear vertical normal mode decomposition. Our results show that the SWTIO is generally dominated by the first baroclinic mode motion. As strong downwelling Rossby waves reach the SWTIO, the contribution of the second baroclinic mode motion in this region can be increased mainly because of the reduction in the vertical stratification of the upper layer above thermocline, and the enhancement in the vertical stratification of the lower layer under thermocline also contributes to it. The vertical displacement of each isothermal is enlarged and the thermal structure of the upper level is modulated, which is indicative of strong vertical mixing. However, the cold Rossby waves increase the vertical stratification of the upper level, restricting the variability related to the second baroclinic mode. On the other hand, during decaying phase of warm Rossby waves, Ekman upwelling and advection processes associated with the surface cyclonic wind circulation can restrain the downwelling processes, carrying the relatively colder water to the near-surface, which results in an out-of-phase phenomenon between sea surface temperature anomaly (SSTA) and sea surface height anomaly (SSHA) in the SWTIO.

Two modes of dipole events in tropical Indian Ocean

By analyzing the distributions of subsurface temperature and the surface wind stress anomalies in the tropical Pacific and Indian Oceans during the Indian Ocean Dipole (IOD) events, two major modes of the IOD and their formation mechanisms are revealed. (1) The subsurface temperature anomaly (STA) in the tropical Indian Ocean during the IOD events can be described as a "<" -shaped and west-east-oriented dipole pattern; in the east side of the "<" pattern, a notable tongue-like STA extends westward along the equator in the tropical eastern Indian Ocean; while in the west side of the "<" pattern, the STA has opposite sign with two centers (the southern one is stronger than the northern one in intensity) being of rough symmetry about the equator in the tropical mid-western Indian Ocean. (2) The IOD events are composed of two modes, which have similar spatial pattern but different temporal variabilities due to the large scale air-sea interactions within two independent systems. The first mode of the IOD event originates from the air-sea interaction on a scale of the tropical Pacific-Indian Ocean and coexists with ENSO. The second mode originates from the air-sea interaction on a scale of the tropical Indian Ocean and is closely associated with changes in the position and intensity of the Mascarene high pressure. The strong IOD event occurs when the two modes are in phase, and the IOD event weakens or disappears when the two modes are out of phase. Besides, the IOD events are normally strong when either of the two modes is strong. (3) The IOD event is caused by the abnormal wind stress forcing over the tropical Indian Ocean, which results in vertical transports, leading to the upwelling and pileup of seawater. This is the main dynamic processes resulting in the STA. When the anomalous easterly exists over the equatorial Indian Ocean, the cold waters upwell in the tropical eastern Indian Ocean while the warm waters pileup in the tropical western Indian Ocean, hence the thermocline in the tropical Indian Ocean is shallowed in the east and deepened in the west. The off-equator component due to the Coriolis force in the equatorial area causes the upwelling of cold waters and the shallowing of the equatorial India Ocean thermocline. On the other hand, the anomalous anticyclonic circulations and their curl fields located on both sides of the equator, cause the pileup of warm waters in the central area of their curl fields and the deepening of the equatorial Indian Ocean thermocline off the equator. The above three factors lead to the occurrence of positive phase IOD events. When anomalous westerly dominates over the tropical Indian Ocean, the dynamic processes are reversed, and the negative-phase IOD event occurs.

Long-Wave Dynamics of Sea Level Variations during Indian Ocean Dipole Events

Long-wave dynamics of the interannual variations of the equatorial Indian Ocean circulation are studied using an ocean general circulation model forced by the assimilated surface winds and heat flux of the European Centre for Medium-Range Weather Forecasts. The simulation has reproduced the sea level anomalies of the Ocean Topography Experiment (TOPEX)/Poseidon altimeter observations well. The equatorial Kelvin and Rossby waves decomposed from the model simulation show that western boundary reflections provide important negative feedbacks to the evolution of the upwelling currents off the Java coast during Indian Ocean dipole (IOD) events. Two downwelling Kelvin wave pulses are generated at the western boundary during IOD events: the first is reflected from the equatorial Rossby waves and the second from the off-equatorial Rossby waves in the southern Indian Ocean. The upwelling in the eastern basin during the 1997-98 IOD event is weakened by the first Kelvin wave pulse and terminated by the second. In comparison, the upwelling during the 1994 IOD event is terminated by the first Kelvin wave pulse because the southeasterly winds off the Java coast are weak at the end of 1994. The atmospheric intraseasonal forcing, which plays an important role in inducing Java upwelling during the early stage of an IOD event, is found to play a minor role in terminating the upwelling off the Java coast because the intraseasonal winds are either weak or absent during the IOD mature phase. The equatorial wave analyses suggest that the upwelling off the Java coast during IOD events is terminated primarily by western boundary reflections.