Generalization bounds for non-stationary mixing processes

This paper presents the first generalization bounds for time series prediction with a non-stationary mixing stochastic process. We prove Rademacher complexity learning bounds for both average-path generalization with non-stationary \(\beta \)-mixing processes and path-dependent generalization with non-stationary \(\phi \)-mixing processes. Our guarantees are expressed in terms of \(\beta \)- or \(\phi \)-mixing coefficients and a natural measure of discrepancy between training and target distributions. They admit as special cases previous Rademacher complexity bounds for non-i.i.d. stationary distributions, for independent but not identically distributed random variables, or for the i.i.d. case. We show that, using a new sub-sample selection technique we introduce, our bounds can be tightened under the natural assumption of asymptotically stationary stochastic processes. We also prove that fast learning rates can be achieved by extending existing local Rademacher complexity analyses to the non-i.i.d. setting. We conclude the paper by providing generalization bounds for learning with unbounded losses and non-i.i.d. data.     

Space-borne study of seasonal,multi-year,and decadal phytoplankton dynamics in the Bay of Biscay

Seasonal and inter-annual variations in phytoplankton community abundance in the Bay of Biscay are studied. Preliminarily processed by the National Aeronautics and Space Administration (NASA) to yield normalized water-leaving radiance and the top-of-the-atmosphere solar radiance, Sea-viewing Wide Field-of-View Sensor (SeaWiFS), Moderate Resolution Imaging Spectroradiometer (MODIS), and Coastal Zone Color Scanner (CZCS) data are further supplied to our dedicated retrieval algorithms to infer the sought for parameters. By applying the National Oceanic and Atmospheric Administration's (NOAA's) Advanced Very High Resolution Radiometer (AVHRR) data, the surface reflection coefficient in the only band in the visible spectrum is derived and employed for analysis. Decadal bridged time series of variations of diatom-dominated phytoplankton and green dinoflagellate Lepidodinium chlorophorum within the shelf zone and the coccolithophore Emiliania huxleyi in the pelagic area of the Bay are documented and analysed in terms of impacts of some biogeochemical and geophysical forcing factors.

It is shown that in the shelf zone of the Bay, the diatom-dominated phytoplankton community variations are predominantly controlled by river discharge variations, by water column stratification conditions (forming in winter–early spring), and by wind action (resulting in such phenomena as up-wellings and sediment re-suspension).

Satellite data indicate that while in river deltas and adjoining waters the L. chlorophorum blooming events occur annually, in the Iroise Sea and near the Bailiwick of Guernsey, they happen irregularly. It is thought that such an irregular pattern, possibly, arises from L. chlorophorum competing with other phytoplankton species for nutrients.

E miliania huxleyi blooms are found to occur nearly every year in the northern part of the Bay, whereas in the central area, this phenomenon occurs very irregularly. Satellite data indicate that variations in the water chemistry (variations in the nitrogen : phosphorus ratio due to preceding blooms of diatoms), and the incident irradiance level (degree of cloudiness), are important factors controlling the occurrence of E. huxleyi blooming in the central part of the Bay. Covering a 30 year period, the bridged data from CZCS, AVHRR, SeaWiFS, and MODIS imply that climate change might be responsible for the observed increase in E. huxleyi blooming events in the Bay since 1979.     

Satellite-derived multi-year trend in primary production in the Arctic Ocean

Spaceborne one month averaged data, predominantly from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) and partly from the Moderate Resolution Imaging Spectroradiometer (MODIS), were used to investigate changes in primary production (PP) by phytoplankton in the Arctic Ocean from 1998 to 2010. Several PP retrieval algorithms were tested against the collected in situ data, and it was shown that the algorithm by Behrenfeld and Falkowski gave the best results (with the coefficient of correlation, r, equal to 0.8 and 0.75, respectively, for the pelagic and shelf zones). Based on the performed test, the Behrenfeld and Falkowski algorithm was further applied for determining both the annual PP in the Arctic and the PP trend over the above-mentioned time period. Results of our analysis indicate that PP in the Arctic has increased by ?15.9% over 13 years (1998–2010). This finding, as well as the absolute annual values of PP remotely quantified in the present study, is at odds with analogous numerical assessments by other workers. These disagreements are thought to be due to differences in the applied methodologies of satellite data processing such as cloud masking and determination of phytoplankton concentration within (1) overcast areas and (2) areas of massive growth of coccolithophores as well as (3) in the shelf zone prone to a significant influence of land and river run-off.     

Ultrasonic analysis of rennet-induced pre-gelation and gelation processes in milk

Dynamics of micro-structural changes in milk during the renneting process were analysed using high-resolution ultrasonic spectroscopy in combination with dynamic rheology and NIR transmission measurements. Two independent ultrasonic parameters, velocity and attenuation were measured in the frequency range 2 to 15 MHz, as a function of time after addition of rennet to milk. The results show an initial decrease of 20 nm for the average diameter of micelles caused by hydrolysis of the kappa-casein 'hairy' layer followed by an aggregation of the micelles into small clusters (effective aggregation number of 3) and then formation of the gel structure. It was found that evolution of ultrasonic attenuation in the renneting process could well be described by the scattering of the ultrasonic waves on aggregates. The evolution of ultrasonic velocity is well described by the scattering theory but deviates from the predicted curve at the gelation stage of the process, which shows the difference in propagation of ultrasonic waves in a gel structure compared with dispersions. Overall, we found high-resolution ultrasonic spectroscopy to be a powerful tool for analysis of microscopic processes in the formation of milk gel. It allows the characterisation of the pre-gelation processes, such as hydrolysis and aggregation, and the initial stages in the formation of the gel network as well as monitoring of the microscopic evolution in the gel at the post-gelation stage.     

Pre-Emptive Scheduling Problems with Controllable Processing Times

We consider a range of single machine and identical parallel machine pre-emptive scheduling models with controllable processing times. For each model we study a single criterion problem to minimize the compression cost of the processing times subject to the constraint that all due dates should be met. We demonstrate that each single criterion problem can be formulated in terms of minimizing a linear function over a polymatroid, and this justifies the greedy approach to its solution. A unified technique allows us to develop fast algorithms for solving both single criterion problems and bicriteria counterparts.     

Fabrication of highly homogeneous Al‐doped TiO2 nanotubes by nanolamination of atomic layer deposition

Conformal parallel arrays of Al‐doped TiO₂ nanotubes were fabricated by atomic layer deposition. TiO₂/Al₂O₃ bilayered shells were grown on a polycarbonate template by various cyclic sequences of TiO₂ and Al₂O₃. The doping level of Al could be tuned by the fraction of cycle number of Al₂O₃. From the depth profiles measured by second ion mass spectrometry, Al is uniformly distributed across the thickness, which is also supported by the analyses of X‐ray diffraction, X‐ray photoelectron spectroscopy, and Raman spectroscopy. A uniform bulk solubility of ~7 at.% and the surface concentration of ~18 at.% were observed with the cycle ratio of Al₂O₃: TiO₂ at 0.04.     

Effect of Residual Solvent on Physicochemical Properties of Poly(Phenylene Isophtalamide) Membrane

The influence of a plasticizer in the form of the residual solvent dimethylacetamide on pervaporation and sorption properties of membranes based on poly(phenylene isophtalamide) (PA) was investigated. To analyze the influence of the plasticizer on membrane transport properties, pervaporation of binary water–ethyl acetate mixtures was studied. The method of sorption calorimetry was used to investigate water sorption properties of the studied membranes and PA powder. Moreover, to characterize the PA membranes, contact angle measurements and thermogravimetric analysis (TGA) were applied. It is shown that the presence of a residual organic solvent significantly changes the sorption and transport characteristics of the membranes. The residual organic solvent increases the amount of water absorbed by the polymer membrane. In pervaporation of water/ethyl acetate mixtures, the presence of the residual solvent dimethylacetamide makes the membrane more permeable but less selective for water separation.     

Comparative study of initial stages of copper immersion deposition on bulk and porous silicon

Initial stages of Cu immersion deposition in the presence of hydrofluoric acid on bulk and porous silicon were studied. Cu was found to deposit both on bulk and porous silicon as a layer of nanoparticles which grew according to the Volmer-Weber mechanism. It was revealed that at the initial stages of immersion deposition, Cu nanoparticles consisted of crystals with a maximum size of 10 nm and inherited the orientation of the original silicon substrate. Deposited Cu nanoparticles were found to be partially oxidized to Cu₂O while CuO was not detected for all samples. In contrast to porous silicon, the crystal orientation of the original silicon substrate significantly affected the sizes, density, and oxidation level of Cu nanoparticles deposited on bulk silicon.     

SERS substrates formed by gold nanorods deposited on colloidal silica films

We describe a new approach to the fabrication of surface-enhanced Raman scattering (SERS) substrates using gold nanorod (GNR) nanopowders to prepare concentrated GNR sols, followed by their deposition on an opal-like photonic crystal (OPC) film formed on a silicon wafer. For comparative experiments, we also prepared GNR assemblies on plain silicon wafers. GNR-OPC substrates combine the increased specific surface, owing to the multilayer silicon nanosphere structure, and various spatial GNR configurations, including those with possible plasmonic hot spots. We demonstrate here the existence of the optimal OPC thickness and GNR deposition density for the maximal SERS effect. All other things being equal, the analytical integral SERS enhancement of the GNR-OPC substrates is higher than that of the thick, randomly oriented GNR assemblies on plain silicon wafers. Several ways to further optimize the strategy suggested are discussed.     

Stretching calorimetry studies of poly(ε-caprolactone)/organoclay nanocomposites

Poly(ε-caprolactone)/montmorillonite nanocomposites, prepared by bulk ring-opening polymerization of ε-caprolactone in the presence of tin(II) 2-ethylhexanoate as a catalyst and hydroxy groups of ammonium tethers of organoclay acting as chain-growth centers, were characterized by the stretching calorimetry technique.In the range of elastic (reversible) deformations, experimental Young's moduli and linear thermal expansivities of both the intercalated and exfoliated nanocomposites could be fitted to the predictions of Chow's model assuming the apparent aspect ratio of clay nanoparticles of the order of 10. In the range of anelastic deformations, the values of the internal energy increments ΔU/m for the intercalated nanocomposite lay somewhat below those for the corresponding neat polymer, while the ΔU/m vs. ε plots for the exfoliated nanocomposite and for the corresponding neat polymer were nearly identical both in shape and in absolute values over the entire range of strains, up to the breaking strains ε_b. These data suggested higher and lower probabilities of interfacial debonding in the range of anelastic deformations for intercalated and exfoliated samples, respectively.