Residual variance estimation in machine learning

The problem of residual variance estimation consists of estimating the best possible generalization error obtainable by any model based on a finite sample of data. Even though it is a natural generalization of linear correlation, residual variance estimation in its general form has attracted relatively little attention in machine learning.In this paper, we examine four different residual variance estimators and analyze their properties both theoretically and experimentally to understand better their applicability in machine learning problems. The theoretical treatment differs from previous work by being based on a general formulation of the problem covering also heteroscedastic noise in contrary to previous work, which concentrates on homoscedastic and additive noise.In the second part of the paper, we demonstrate practical applications in input and model structure selection. The experimental results show that using residual variance estimators in these tasks gives good results often with a reduced computational complexity, while the nearest neighbor estimators are simple and easy to implement.     

Modeling realistic hybrid flexible flowshop scheduling problems

This paper aims to contribute to the recent research efforts to bridge the gap between the theory and the practice of scheduling by modelizing a realistic manufacturing environment and analyzing the effect of the inclusion of several characteristics in the problem formulation. There are several constraints and characteristics that affect the scheduling operations at companies. While these constraints are many times tackled in the literature, they are seldom considered together inside the same problem formulation. We propose a formulation along with a mixed integer modelization and some heuristics for the problem of scheduling n jobs on m stages where at each stage we have a known number of unrelated machines. The jobs might skip stages and, therefore, we have what we call a hybrid flexible flowshop problem. We also consider per machine sequence-dependent setup times which can be anticipatory and non-anticipatory along with machine lags, release dates for machines, machine eligibility and precedence relationships among jobs. Manufacturing environments like this appear in sectors like food processing, ceramic tile manufacturing and several others. The optimization criterion considered is the minimization of the makespan. The MIP model and the heuristics proposed are tested against a comprehensive benchmark and the results evaluated by advanced statistical tools that make use of decision trees and experimental designs. The results allow us to identify the constraints that increase the difficulty.     

A Biodegradable Multifunctional Graphene Oxide Platform for Targeted Cancer Therapy

The design of multifunctional materials able to both selectively deliver a drug into cells in a targeted manner and display an enhanced propensity for biodegradation is an important goal. Here, graphene oxide (GO) is functionalized with the chemotactic peptide N‐formyl‐methionyl‐leucyl‐phenylalanine (fMLP) known to interact with the formyl peptide receptor, which is expressed in different cancer cells, including cervical carcinoma cells. This study highlights the ability of GOfMLP for targeted drug delivery and cancer cell killing and the subsequent degradation capacity of the hybrid. Biodegradation is assessed via Raman spectroscopy and transmission electron microscopy. The results show that GOfMLP is susceptible to faster myeloperoxidase‐mediated degradation. The hybrid material, but not GO, is capable of inducing neutrophil degranulation with subsequent degradation, being the first study showing inducible neutrophil degradation by the nanomaterial itself with no prior activation of the cells. In addition, confocal imaging and flow cytometry using HeLa cells demonstrate that GOfMLP is able to deliver the chemotherapeutic agent doxorubicin faster into cells, inducing higher levels of apoptosis, when compared to nonfunctionalized GO. The results reveal that GOfMLP is a promising carrier able to efficiently deliver anticancer drugs, being endowed with the ability to induce its own biodegradation.     

Preparation and characterization of three pure magnesium vanadate phases as catalysts for selective oxidation of propane to propene

Three magnesium vanadate phases, i.e., MgV₂O₆ (metavanadate), -Mg₂V₂O₇ (pyrovana-date) and Mg_s V₂O₈ (orthovanadate), have been successfully prepared with high purity by the citrate method at a relatively low temperature (550°C). FT-IR, LRS, XRD and SEM techniques have been used to characterize these vanadate phases. The effect of calcination temperature has also been investigated. It was found that the particle size and morphology of the MgV₂O₆ phase, which is a function of calcination temperature, appear to have a strong effect on the infrared spectra. Furthermore, the catalytic properties of the three phases were examined in the oxidative dehydrogenation of propane. The propene selectivity follows the order: -Mg₂V₂O₇ > Mg₃ Vg₂O₈ > MgV₂O₆, which is consistent with their redox properties. This fact suggests that there is some correlation between the catalytic and redox properties of these magnesium vanadate phases.     

Engineering Transport in Manganites by Tuning Local Nonstoichiometry in Grain Boundaries

Interface‐dominated materials such as nanocrystalline thin films have emerged as an enthralling class of materials able to engineer functional properties of transition metal oxides widely used in energy and information technologies. In particular, it has been proven that strain‐induced defects in grain boundaries of manganites deeply impact their functional properties by boosting their oxygen mass transport while abating their electronic and magnetic order. In this work, the origin of these dramatic changes is correlated for the first time with strong modifications of the anionic and cationic composition in the vicinity of strained grain boundary regions. We are also able to alter the grain boundary composition by tuning the overall cationic content in the films, which represents a new and powerful tool, beyond the classical space charge layer effect, for engineering electronic and mass transport properties of metal oxide thin films useful for a collection of relevant solid‐state devices.     

Soft,Skin‐Interfaced Microfluidic Systems with Passive Galvanic Stopwatches for Precise Chronometric Sampling of Sweat

Comprehensive analysis of sweat chemistry provides noninvasive health monitoring capabilities that complement established biophysical measurements such as heart rate, blood oxygenation, and body temperature. Recent developments in skin‐integrated soft microfluidic systems address many challenges associated with standard technologies in sweat collection and analysis. However, recording of time‐dependent variations in sweat composition requires bulky electronic systems and power sources, thereby constraining form factor, cost, and modes of use. Here, presented are unconventional design concepts, materials, and device operation principles that address this challenge. Flexible galvanic cells embedded within skin‐interfaced microfluidics with passive valves serve as sweat‐activated “stopwatches” that record temporal information associated with collection of discrete microliter volumes of sweat. The result allows for precise measurements of dynamic sweat composition fluctuations using in situ or ex situ analytical techniques. Integrated electronics based on near‐field communication (NFC) protocols or docking stations equipped with standard electronic measurement tools provide means for extracting digital timing results from the stopwatches. Human subject studies of time‐stamped sweat samples by in situ colorimetric methods and ex situ techniques based on inductively coupled plasma mass spectroscopy (ICP‐MS) and chlorodimetry illustrate the ability to quantitatively capture time‐dynamic sweat chemistry in scenarios compatible with field use.     

Considering scheduling and preventive maintenance in the flowshop sequencing problem

The aim of this paper is to propose tools in order to implicitly consider different preventive maintenance policies on machines regarding flowshop problems. These policies are intended to maximize the availability or to keep a minimum level of reliability during the production horizon. It proposes a simple criterion to schedule preventive maintenance operations to the production sequence. This criterion demonstrates the significance of taking into consideration preventive maintenance together with sequencing and the consequences of not doing so. The optimization criterion considered consists in minimizing the makespan of the sequence or _Cmax$C_{\max }$. In total, six adaptations of existing heuristic and metaheuristic methods are evaluated for the consideration of preventive maintenance and they are applied to a set of 7200 instances. The results and experiments carried out indicate that modern Ant Colony and Genetic Algorithms provide very effective solutions for this problem.     

State Variable Implications on Hydraulic State Estimation

Hydraulic State Estimation allows to estimate the most likely state of a water supply network considering multiple sources of information and their associated uncertainty. It is set out as an optimization problem, often addressed according to the Weighted Least Squares criterion. It can be formulated differently depending on the selected set of state variables. This choice is not straightforward and leads to different problem dimensions, time complexities and convergence behavior. All possible essential approaches are gathered in this work and two are identified as the most suitable according to the resulting problem dimensions: a head-based approach and a demand-based approach. The particularities and the formulation according to both sets of state variables are discussed, seeking for efficient implementation. Their time complexity and convergence behavior are compared to draw conclusions that help to identify the most suitable approach for real practice applications.

     

Potential Performance of Hydraulic State Estimation in Water Distribution Networks

Water Resources Management - Hydraulic State Estimation (HSE) is a suitable tool to gain real-time understanding of water supply systems. This technique enables to estimate the most likely...     

Further on the mechanism of the synergy between MoO3 and α-Sb2O4 in the selective oxidation of isobutene to methacrolein: Reconstuction of MoO3 via spillover oxygen

This paper concerns the synergetic effects between a (010) oriented MoO₃ and α-Sb₂O₄ in the oxidation of isobutene to methacrolein at 420°C. The catalysts were prepared by mechanical mixture of the pure oxides prepared separately. Important increases were observed both for the conversion of isobutene and for the selectivity to methacrolein, suggesting that new selective sites have been created. The characterization of the samples before and after the reaction by SEM and XRD shows an important reconstruction of the (010) face of MoO₃ into steps exposing (100) faces, when the reaction is carried out in the presence of α-Sb₂O₄. It is known from literature that the (100) face of MoO₃ performs selectively partial oxidation, while the (010) face is not selective. On the other hand, α-Sb₂O₄ has shown to produce spillover oxygen, which flows onto the MoO₃ surface. It is, therefore, concluded that the increase of selectivity is intimately connected to the reconstruction under the action of spillover oxygen. Spillover oxygen would favour the “selective coordination” of Mo atoms or groups of atoms, typical of (100), at the expense of the “non-selective one”, typical of (010). This result corresponds to the theoretical prediction of the Remote Control theory.