Domain Adaptation in remote sensing: increasing the portability of land-cover classifiers

Among the types of remote sensing acquisitions, optical images are certainly one of the most widely relied upon data sources for Earth observation. They provide detailed measurements of the electromagnetic radiation reflected or emitted by each pixel in the scene. Through a process termed supervised land-cover classification, this allows to automatically yet accurately distinguish objects at the surface of our planet. In this respect, when producing a land-cover map of the surveyed area, the availability of training examples representative of each thematic class is crucial for the success of the classification procedure. However, in real applications, due to several constraints on the sample collection process, labeled pixels are usually scarce. When analyzing an image for which those key samples are unavailable, a viable solution consists in resorting to the ground truth data of other previously acquired images. This option is attractive but several factors such as atmospheric, ground and acquisition conditions can cause radiometric differences between the images, hindering therefore the transfer of knowledge from one image to another. The goal of this Thesis is to supply remote sensing image analysts with suitable processing techniques to ensure a robust portability of the classification models across different images. The ultimate purpose is to map the land-cover classes over large spatial and temporal extents with minimal ground information. To overcome, or simply quantify, the observed shifts in the statistical distribution of the spectra of the materials, we study four approaches issued from the field of machine learning. First, we propose a strategy to intelligently sample the image of interest to collect the labels only in correspondence of the most useful pixels. This iterative routine is based on a constant evaluation of the pertinence to the new image of the initial training data actually belonging to a different image. Second, an approach to reduce the radiometric differences among the images by projecting the respective pixels in a common new data space is presented. We analyze a kernel-based feature extraction framework suited for such problems, showing that, after this relative normalization, the cross-image generalization abilities of a classifier are highly increased. Third, we test a new data-driven measure of distance between probability distributions to assess the distortions caused by differences in the acquisition geometry affecting series of multi-angle images. Also, we gauge the portability of classification models through the sequences. In both exercises, the efficacy of classic physically- and statistically-based normalization methods is discussed. Finally, we explore a new family of approaches based on sparse representations of the samples to reciprocally convert the data space of two images. The projection function bridging the images allows a synthesis of new pixels with more similar characteristics ultimately facilitating the land-cover mapping across images.

A time-domain method for the analysis of thermal impedance response preserving the convolution form

The study of the thermal behavior of complex packages as multichip modules (MCM¿s) is usually carried out by measuring the so-called thermal impedance response, that is: the transient temperature after a power step. From the analysis of this signal, the thermal frequency response can be estimated, and consequently, compact thermal models may be extracted. We present a method to obtain an estimate of the time constant distribution underlying the observed transient. The method is based on an iterative deconvolution that produces an approximation to the time constant spectrum while preserving a convenient convolution form. This method is applied to the obtained thermal response of a microstructure as analyzed by finite element method as well as to the measured thermal response of a transistor array integrated circuit (IC) in a SMD package.

c-Jun N-terminal kinase binding domain-dependent phosphorylation of mitogen-activated protein kinase kinase 4 and mitogen-activated protein kinase kinase 7 and balancing cross-talk between c-Jun N-terminal kinase and extracellular signal-regulated kinase pathways in cortical neurons

The c-Jun N-terminal kinase (JNK) is a mitogen-activated protein kinase (MAPK) activated by stress-signals and involved in many different diseases. Previous results proved the powerful effect of the cell permeable peptide inhibitor d-JNKI1 (d-retro-inverso form of c-Jun N-terminal kinase-inhibitor) against neuronal death in CNS diseases, but the precise features of this neuroprotection remain unclear. We here performed cell-free and in vitro experiments for a deeper characterization of d-JNKI1 features in physiological conditions. This peptide works by preventing JNK interaction with its c-Jun N-terminal kinase-binding domain (JBD) dependent targets. We here focused on the two JNK upstream MAPKKs, mitogen-activated protein kinase kinase 4 (MKK4) and mitogen-activated protein kinase kinase 7 (MKK7), because they contain a JBD homology domain. We proved that d-JNKI1 prevents MKK4 and MKK7 activity in cell-free and in vitro experiments: these MAPKK could be considered not only activators but also substrates of JNK. This means that d-JNKI1 can interrupt downstream but also upstream events along the JNK cascade, highlighting a new remarkable feature of this peptide. We also showed the lack of any direct effect of the peptide on p38, MEK1, and extracellular signal-regulated kinase (ERK) in cell free, while in rat primary cortical neurons JNK inhibition activates the MEK1-ERK-Ets1/c-Fos cascade. JNK inhibition induces a compensatory effect and leads to ERK activation via MEK1, resulting in an activation of the survival pathway-(MEK1/ERK) as a consequence of the death pathway-(JNK) inhibition. This study should hold as an important step to clarify the strong neuroprotective effect of d-JNKI1.

Option valuation as an expectation in the complex domain: The Black-Scholes case

It is very well known that the first succesful valuation of a stock option was done by solving a deterministic partial differential equation (PDE) of the parabolic type with some complementary conditions specific for the option. In this approach, the randomness in the option value process is eliminated through a no-arbitrage argument. An alternative approach is to construct a replicating portfolio for the option. From this viewpoint the payoff function for the option is a random process which, under a new probabilistic measure, turns out to be of a special type, a martingale. Accordingly, the value of the replicating portfolio (equivalently, of the option) is calculated as an expectation, with respect to this new measure, of the discounted value of the payoff function. Since the expectation is, by definition, an integral, its calculation can be made simpler by resorting to powerful methods already available in the theory of analytic functions. In this paper we use precisely two of those techniques to find the well-known value of a European call

Domain growth in binary mixtures at low temperatures

We have studied domain growth during spinodal decomposition at low temperatures. We have performed a numerical integration of the deterministic time-dependent Ginzburg-Landau equation with a variable, concentration-dependent diffusion coefficient. The form of the pair-correlation function and the structure function are independent of temperature but the dynamics is slower at low temperature. A crossover between interfacial diffusion and bulk diffusion mechanisms is observed in the behavior of the characteristic domain size. This effect is explained theoretically in terms of an equation of motion for the interface.

Effects of domain morphology in phase-separation dynamics at low temperature

We present numerical results of the deterministic Ginzburg-Landau equation with a concentration-dependent diffusion coefficient, for different values of the volume fraction phi of the minority component. The morphology of the domains affects the dynamics of phase separation. The effective growth exponents, but not the scaled functions, are found to be temperature dependent.

Fluctuations in domain growth: Ginzburg-Landau equations with multiplicative noise

Ginzburg-Landau equations with multiplicative noise are considered, to study the effects of fluctuations in domain growth. The equations are derived from a coarse-grained methodology and expressions for the resulting concentration-dependent diffusion coefficients are proposed. The multiplicative noise gives contributions to the Cahn-Hilliard linear-stability analysis. In particular, it introduces a delay in the domain-growth dynamics.

Numerical algorithm for Ginzburg-Landau equations with multiplicative noise: Application to domain growth

We consider stochastic partial differential equations with multiplicative noise. We derive an algorithm for the computer simulation of these equations. The algorithm is applied to study domain growth of a model with a conserved order parameter. The numerical results corroborate previous analytical predictions obtained by linear analysis.

Front and domain growth in the presence of gravity

Front and domain growth of a binary mixture in the presence of a gravitational field is studied. The interplay of bulk- and surface-diffusion mechanisms is analyzed. An equation for the evolution of interfaces is derived from a time-dependent Ginzburg-Landau equation with a concentration-dependent diffusion coefficient. Scaling arguments on this equation give the exponents of a power-law growth. Numerical integrations of the Ginzburg-Landau equation corroborate the theoretical analysis.

Monte Carlo study of the relation between vacancy diffusion and domain growth in two-dimensional binary alloys

Domain growth in a two-dimensional binary alloy is studied by means of Monte Carlo simulation of an ABV model. The dynamics consists of exchanges of particles with a small concentration of vacancies. The influence of changing the vacancy concentration and finite-size effects has been analyzed. Features of the vacancy diffusion during domain growth are also studied. The anomalous character of the diffusion due to its correlation with local order is responsible for the obtained fast-growth behavior.

Electrical conductivity of soil irrigated with swine wastewater estimated by time-domain reflectometry¹

Wastewater application to soil is an alternative for fertilization and water reuse. However, particular care must be taken with this practice, since successive wastewater applications can cause soil salinization. Time-domain reflectometry (TDR) allows for the simultaneous and continuous monitoring of both soil water content and apparent electrical conductivity and thus for the indirect measurement of the electrical conductivity of the soil solution. This study aimed to evaluate the suitability of TDR for the indirect determination of the electrical conductivity (ECse) of the saturated soil extract by using an empirical equation for the apparatus TDR Trase 6050X1. Disturbed soil samples saturated with swine wastewater were used, at soil proportions of 0, 0.45, 0.90, 1.80, 2.70, and 3.60 m³ m-3. The probes were equipped with three handmade 0.20 cm long rods. The fit of the empirical model that associated the TDR measured values of electrical conductivity (EC TDR) to ECse was excellent, indicating this approach as suitable for the determination of electrical conductivity of the soil solution.

The emerging importance of the SPX domain-containing proteins in phosphate homeostasis

Plant growth and development are strongly influenced by the availability of nutrients in the soil solution. Among them, phosphorus (P) is one of the most essential and most limiting macro-elements for plants. In the environment, plants are often confronted with P starvation as a result of extremely low concentrations of soluble inorganic phosphate (Pi) in the soil. To cope with these conditions, plants have developed a wide spectrum of mechanisms aimed at increasing P use efficiency. At the molecular level, recent studies have shown that several proteins carrying the SPX domain are essential for maintaining Pi homeostasis in plants. The SPX domain is found in numerous eukaryotic proteins, including several proteins from the yeast PHO regulon, involved in maintaining Pi homeostasis. In plants, proteins harboring the SPX domain are classified into four families based on the presence of additional domains in their structure, namely the SPX, SPX-EXS, SPX-MFS and SPX-RING families. In this review, we highlight the recent findings regarding the key roles of the proteins containing the SPX domain in phosphate signaling, as well as providing further research directions in order to improve our knowledge on P nutrition in plants, thus enabling the generation of plants with better P use efficiency.

Dynamics of domain walls in pattern formation under traveling-wave forcing

We study dynamics of domain walls in pattern forming systems that are externally forced by a moving space-periodic modulation close to 2:1 spatial resonance. The motion of the forcing induces nongradient dynamics, while the wave number mismatch breaks explicitly the chiral symmetry of the domain walls. The combination of both effects yields an imperfect nonequilibrium Ising-Bloch bifurcation, where all kinks (including the Ising-like one) drift. Kink velocities and interactions are studied within the generic amplitude equation. For nonzero mismatch, a transition to traveling bound kink-antikink pairs and chaotic wave trains occurs.