A self-adaptive synthetic over-sampling technique for imbalanced classification

Traditionally, in supervised machine learning, (a significant) part of the available data (usually 50%-80%) is used for training and the rest—for validation. In many problems, however, the data are highly imbalanced in regard to different classes or does not have good coverage of the feasible data space which, in turn, creates problems in validation and usage phase. In this paper, we propose a technique for synthesizing feasible and likely data to help balance the classes as well as to boost the performance in terms of confusion matrix as well as overall. The idea, in a nutshell, is to synthesize data samples in close vicinity to the actual data samples specifically for the less represented (minority) classes. This has also implications to the so-called fairness of machine learning. In this paper, we propose a specific method for synthesizing data in a way to balance the classes and boost the performance, especially of the minority classes. It is generic and can be applied to different base algorithms, for example, support vector machines, k-nearest neighbour classifiers deep neural, rule-based classifiers, decision trees, and so forth. The results demonstrated that (a) a significantly more balanced (and fair) classification results can be achieved and (b) that the overall performance as well as the performance per class measured by confusion matrix can be boosted. In addition, this approach can be very valuable for the cases when the number of actual available labelled data is small which itself is one of the problems of the contemporary machine learning.     

The properties of a spinel copper cobaltite prepared at low temperatures and normal pressure

The properties of a spinel copper cobaltite obtained by thermal decomposition of mixed nitrates of both metals at 350°C and normal pressure are described. It has been shown that the maximum copper content (x) in the spinel Cu_xCo_3?xO₄ does not exceed 0.9. When the copper amount in the initial mixture is higher a second phase CuO is formed. The copper ions are distributed over the A- and the B-sites of the spinel lattice, i.e. the spinel is partially inverted. Such a model explains well the data obtained about the lattice parameter, IR-absorption and the magnetic susceptibility of the samples with different x.     

Evidence of flexoelectricity in polymer-dispersed liquid crystals

The linear electro-optic response of polymer-dispersed liquid crystal films has been studied. Micrometer-sized droplets of a cyanobiphenyl mixture (E7) were formed in a Norland Optical Adhesive 65 photopolymer by photopolymerization-induced phase separation. Samples with various average nematic droplet sizes were systematically investigated. The temperature dependence of the first-harmonic electro-optic spectra was obtained. Peculiar peaks and double peaks in the phase spectra were found at temperatures close to the clearing point. Samples with narrow distributions of droplet sizes were characterized by a double peak of the amplitude spectra in the low-frequency range. A flexoelectric origin of the first-harmonic spectra in confined nematic systems is discussed.     

Robust self-assembly of graphs

Self-assembly is a process in which small building blocks interact autonomously to form larger structures. A recently studied model of self-assembly is the Accretive Graph Assembly Model whereby an edge-weighted graph is assembled one vertex at a time starting from a designated seed vertex. The weight of an edge specifies the magnitude of attraction (positive weight) or repulsion (negative weight) between adjacent vertices. It is feasible to add a vertex to the assembly if the total attraction minus repulsion of the already built neighbors exceeds a certain threshold, called the assembly temperature. This model naturally generalizes the extensively studied Tile Assembly Model. A natural question in graph self-assembly is to determine whether or not there exists a sequence of feasible vertex additions to realize the entire graph. However, even when it is feasible to realize the assembly, not much can be inferred about its likelihood of realization in practice due to the uncontrolled nature of the self-assembly process. Motivated by this, we introduce the robust self-assembly problem where the goal is to determine if every possible sequence of feasible vertex additions leads to the completion of the assembly. We show that the robust self-assembly problem is co-NP-complete even on planar graphs with two distinct edge weights. We then examine the tractability of the robust self-assembly problem on a natural subclass of planar graphs, namely grid graphs. We identify structural conditions that determine whether or not a grid graph can be robustly self-assembled, and give poly-time algorithms to determine this for several interesting cases of the problem. Finally, we also show that the problem of counting the number of feasible orderings that lead to the completion of an assembly is #P-complete.     

The Network as a Storage Device: Dynamic Routing with Bounded Buffers

We study dynamic routing in store-and-forward packet networks where each network link has bounded buffer capacity for receiving incoming packets and is capable of transmitting a fixed number of packets per unit of time. At any moment in time, packets are injected at various network nodes with each packet specifying its destination node. The goal is to maximize the throughput, defined as the number of packets delivered to their destinations. In this paper, we make some progress on throughput maximization in various network topologies. Let n and m denote the number of nodes and links in the network, respectively. For line networks, we show that Nearest-to-Go (NTG), a natural distributed greedy algorithm, is -competitive, essentially matching a known lower bound on the performance of any greedy algorithm. We also show that if we allow the online routing algorithm to make centralized decisions, there is a randomized polylog(n)-competitive algorithm for line networks as well as for rooted tree networks, where each packet is destined for the root of the tree. For grid graphs, we show that NTG has a competitive ratio of while no greedy algorithm can achieve a ratio better than . Finally, for arbitrary network topologies, we show that NTG is -competitive, improving upon an earlier bound of O(mn). An extended abstract appeared in the Proceedings of the 8th Workshop on Approximation Algorithms for Combinatorial Optimization Problems, APPROX 2005, Berkeley, CA, USA, pp. 1–13, Lecture Notes in Computer Science, vol. 1741, Springer, Berlin. S. Angelov is supported in part by NSF Career Award CCR-0093117, NSF Award ITR 0205456 and NIGMS Award 1-P20-GM-6912-1. S. Khanna is supported in part by an NSF Career Award CCR-0093117, NSF Award CCF-0429836, and a US-Israel Binational Science Foundation Grant. K. Kunal is supported in part by an NSF Career Award CCR-0093117 and NSF Award CCF-0429836.     

Patterns and tools for business process monitoring customization

In a cross-organizational service-based process provisioning scenario, one provider is likely to execute a given business process to serve several customers. Each customer may hold different expectations about the way this process can be monitored. We present a solution allowing the provider to support the requirements of different customers on the monitoring of a given process, i.e., offering them the opportunity to customize the way a process will be monitored. We propose a multi-dimensional classification model of patterns for process monitoring and rules to compose the patterns to design customized monitoring infrastructures. The fit for purpose of the patterns is evaluated empirically, whereas the feasibility of our solution is demonstrated by a tool supporting process monitoring customization adhering to our pattern design and composition methodology.     

Fuzzy systems design: direct and indirect approaches

A systematic classification of the data-driven approaches for design of fuzzy systems is given in the paper. The possible ways to solve this modelling and identification problem are classified on the basis of the optimisation techniques used for this purpose. One algorithm for each of the two basic categories of design methods is presented and its advantages and disadvantages are discussed. Both types of algorithms are self-learning and do not require interaction during the process of fuzzy model design. They perform adaptation of both the fuzzy model structure (rule-base) and the parameters. The indirect approach exploits the dual nature of Takagi-Sugeno (TS) models and is based on recently introduced recursive clustering combined with Kalman filtering-based procedure for recursive estimation of the parameter of the local sub-models. Both algorithms result in finding compact and transparent fuzzy models. The direct approach solves the optimisation problem directly, while the indirect one decomposes the original problem into on-line clustering and recursive estimation problems and finds a sub-optimal solution in real-time. The later one is computationally very efficient and has a range of potential applications in real-time process control, moving images recognition, autonomous systems design etc. It is extended in this paper for the case of multi-input–multi-output (MIMO systems). Both approaches have been tested with real data from an engineering process.     

Scale and translation invariant minimum average correlation energy filter

A new type, to our knowledge, of scale and translation invariant correlation filter is described. Its form in polar coordinates generalizes the forms of the one-decomposition-term filters. That is combined with the minimum-average-correlation-energy optimization method to suppress the sidelobes and achieve correlation peaks. A theoretical analysis as well as a detailed explanation of the computational procedure is provided. The concept is tested on five interferometric images of 256 gray levels without preprocessing. The computed filter gives correlation peaks for reference image scales in the range of 0.3/4.0 (minimal scale: maximal scale = 1:13). The discrimination ability of the filter is investigated-no false peak occurred. The filter works in the background so that there is no need for input image segmentation. A method for extension is described to calculate a filter for more than one reference image.     

Preparation and properties of iron molybdates. II - Magnetochemistry of crystalline phases obtained from precursor precipitates with different compositions

The crystalline phases prepared by calcination at 350°C of amorphous iron molybdate precipitates with different chemical composition have been studied by magnetic susceptibility and EPR measurements. The magnetic susceptibility parameters and peak-to-peak resonance linewidth are defined for the stoichiometric Fe₂(MoO₄)₃.At MotoFe ratios in the precursors higher than 1.5 the crystal products contain lattice defects changing the values of the intensive magnetic parameters.     

Fast Batch to Continuous Transposition: Application to the Extraction of Andrographolide from Plants

A fast development method for batch to continuous process transposition is proposed. This method is based on transient regime experiment analyses and is applied to a solid‐liquid extraction. The application under consideration is the extraction of an active principle from a plant in a non‐sinusoidal pulsed column. Typically, the proposed signal is composed of two different periods: firstly, a classical sinusoidal pulsation step is used to mix the liquid and solid phases in the active part of the column and allow an optimal mass transfer and, secondly, an impulsion phase, used generally for the transport of solids. The extraction is carried out in a disc and doughnut column of 54 mm diameter and 3.5 m height. Liquid and solid are flowing co‐currently and downwardly. This technological improvement has been implemented to solve the difficulties due to the significant heterogeneity of the matter: one part tends to float and other to sink, which always leads to a definitive flooding in classical operations. The effects of the solid flow rate and the solvent characteristics on the hydrodynamic behavior of the column are studied. The mean residence time and the total solid holdup are calculated by using a transient regime mass balance on the experimental results. These experiments allow the identification and quantification of opposite effects of the operating parameters. Mass transfer experiments have been performed and the results fit calculated values obtained by coupling the hydrodynamic and batch extraction results. Despite the simplifications made, this validates the fast development method proposed to help batch to continuous transposition.