Efficient Monte Carlo methods for multi-dimensional learning with classifier chains

Multi-dimensional classification (MDC) is the supervised learning problem where an instance is associated with multiple classes, rather than with a single class, as in traditional classification problems. Since these classes are often strongly correlated, modeling the dependencies between them allows MDC methods to improve their performance – at the expense of an increased computational cost. In this paper we focus on the classifier chains (CC) approach for modeling dependencies, one of the most popular and highest-performing methods for multi-label classification (MLC), a particular case of MDC which involves only binary classes (i.e., labels). The original CC algorithm makes a greedy approximation, and is fast but tends to propagate errors along the chain. Here we present novel Monte Carlo schemes, both for finding a good chain sequence and performing efficient inference. Our algorithms remain tractable for high-dimensional data sets and obtain the best predictive performance across several real data sets.

Student academic performance stochastic simulator based on the Monte Carlo method

In this paper, a computer-based tool is developed to analyze student performance along a given curriculum. The proposed software makes use of historical data to compute passing/failing probabilities and simulates future student academic performance based on stochastic programming methods (MonteCarlo) according to the specific university regulations. This allows to compute the academic performance rates for the specific subjects of the curriculum for each semester, as well as the overall rates (the set of subjects in the semester), which are the efficiency rate and the success rate. Additionally, we compute the rates for the Bachelors degree, which are the graduation rate measured as the percentage of students who finish as scheduled or taking an extra year and the efficiency rate (measured as the percentage of credits of the curriculum with respect to the credits really taken). In Spain, these metrics have been defined by the National Quality Evaluation and Accreditation Agency (ANECA). Moreover, the sensitivity of the performance metrics to some of the parameters of the simulator is analyzed using statistical tools (Design of Experiments). The simulator has been adapted to the curriculum characteristics of the Bachelor in Engineering Technologies at the Technical University of Madrid(UPM).

Monte Carlo limit cycle characterization

The fixed point implementation of IIR digital filters usually leads to the appearance of zero-input limit cycles, which degrade the performance of the system. In this paper, we develop an efficient Monte Carlo algorithm to detect and characterize limit cycles in fixed-point IIR digital filters. The proposed approach considers filters formulated in the state space and is valid for any fixed point representation and quantization function. Numerical simulations on several high-order filters, where an exhaustive search is unfeasible, show the effectiveness of the proposed approach.

Monte Carlo simulations of densely-packed athermal polymers in the bulk and under confinement

We review the main results from extensive Monte Carlo (MC) simulations on athermal polymer packings in the bulk and under confinement. By employing the simplest possible model of excluded volume, macromolecules are represented as freely-jointed chains of hard spheres of uniform size. Simulations are carried out in a wide concentration range: from very dilute up to very high volume fractions, reaching the maximally random jammed (MRJ) state. We study how factors like chain length, volume fraction and flexibility of bond lengths affect the structure, shape and size of polymers, their packing efficiency and their phase behaviour (disorder–order transition). In addition, we observe how these properties are affected by confinement realized by flat, impenetrable walls in one dimension. Finally, by mapping the parent polymer chains to primitive paths through direct geometrical algorithms, we analyse the characteristics of the entanglement network as a function of packing density.

A stochastic approximation algorithm with Markov chain Monte-Carlo method for incomplete data estimation problems

We propose a general procedure for solving incomplete data estimation problems. The procedure can be used to find the maximum likelihood estimate or to solve estimating equations in difficult cases such as estimation with the censored or truncated regression model, the nonlinear structural measurement error model, and the random effects model. The procedure is based on the general principle of stochastic approximation and the Markov chain Monte-Carlo method. Applying the theory on adaptive algorithms, we derive conditions under which the proposed procedure converges. Simulation studies also indicate that the proposed procedure consistently converges to the maximum likelihood estimate for the structural measurement error logistic regression model.

Dynamic weighting in Monte Carlo and optimization

Dynamic importance weighting is proposed as a Monte Carlo method that has the capability to sample relevant parts of the configuration space even in the presence of many steep energy minima. The method relies on an additional dynamic variable (the importance weight) to help the system overcome steep barriers. A non-Metropolis theory is developed for the construction of such weighted samplers. Algorithms based on this method are designed for simulation and global optimization tasks arising from multimodal sampling, neural network training, and the traveling salesman problem. Numerical tests on these problems confirm the effectiveness of the method.

Enhanced conformational sampling in Monte Carlo simulations of proteins: application to a constrained peptide.

A Monte Carlo simulation method for globular proteins, called extended-scaled-collective-variable (ESCV) Monte Carlo, is proposed. This method combines two Monte Carlo algorithms known as entropy-sampling and scaled-collective-variable algorithms. Entropy-sampling Monte Carlo is able to sample a large configurational space even in a disordered system that has a large number of potential barriers. In contrast, scaled-collective-variable Monte Carlo provides an efficient sampling for a system whose dynamics is highly cooperative. Because a globular protein is a disordered system whose dynamics is characterized by collective motions, a combination of these two algorithms could provide an optimal Monte Carlo simulation for a globular protein. As a test case, we have carried out an ESCV Monte Carlo simulation for a cell adhesive Arg-Gly-Asp-containing peptide, Lys-Arg-Cys-Arg-Gly-Asp-Cys-Met-Asp, and determined the conformational distribution at 300 K. The peptide contains a disulfide bridge between the two cysteine residues. This bond mimics the strong geometrical constraints that result from a protein's globular nature and give rise to highly cooperative dynamics. Computation results show that the ESCV Monte Carlo was not trapped at any local minimum and that the canonical distribution was correctly determined.

Reconfigurable computing for Monte Carlo simulations: results and prospects of the Janus project

We describe Janus, a massively parallel FPGA-based computer optimized for the simulation of spin glasses, theoretical models for the behavior of glassy materials. FPGAs (as compared to GPUs or many-core processors) provide a complementary approach to massively parallel computing. In particular, our model problem is formulated in terms of binary variables, and floating-point operations can be (almost) completely avoided. The FPGA architecture allows us to run many independent threads with almost no latencies in memory access, thus updating up to 1024 spins per cycle. We describe Janus in detail and we summarize the physics results obtained in four years of operation of this machine; we discuss two types of physics applications: long simulations on very large systems (which try to mimic and provide understanding about the experimental non equilibrium dynamics), and low-temperature equilibrium simulations using an artificial parallel tempering dynamics. The time scale of our non-equilibrium simulations spans eleven orders of magnitude (from picoseconds to a tenth of a second). On the other hand, our equilibrium simulations are unprecedented both because of the low temperatures reached and for the large systems that we have brought to equilibrium. A finite-time scaling ansatz emerges from the detailed comparison of the two sets of simulations. Janus has made it possible to perform spin glass simulations that would take several decades on more conventional architectures. The paper ends with an assessment of the potential of possible future versions of the Janus architecture, based on state-of-the-art technology.

Estudo dos parâmetros dosimétricos de sementes de Iodo-125 desenvolvidas pelo IPEN-CNEN/SP utilizadas em braquiterapia por simulação computacional pelo método de Monte Carlo

As expectativas da Organização Mundial de Saúde para o ano de 2030 são que o número de mortes por câncer seja de aproximadamente 13,2 milhões, evidenciando a elevada parcela desta doença no problema de saúde mundial. Com relação ao câncer de próstata, de acordo com o Instituto Nacional do Câncer, o número de casos diagnosticados no mundo em 2012 foi de aproximadamente 1,1 milhão, enquanto que no Brasil os dados indicam a incidência de 68 mil novos casos. O tratamento deste tipo de neoplasia pode ser realizado com cirurgia (prostatectomia) ou radioterapia. Dentre a radioterapia, podemos destacar a técnica de braquiterapia, a qual consiste na introdução (implante) de pequenas fontes radioativas (sementes) no interior da próstata, onde será entregue um valor elevado de dose no volume de tratamento e baixa dose nos tecidos ao redor. No Brasil, a classe médica estima uma demanda de aproximadamente 8000 sementes/mês, sendo o custo unitário de cada semente de pelo menos U$ 26,00. A Associação Americana de Físicos na Medicina publicou alguns documentos descrevendo quais parâmetros e análises devem ser realizadas para avaliações da distribuição de dose, como por exemplo, os parâmetros Constante de taxa de dose, Função radial e Função de anisotropia. Estes parâmetros podem ser obtidos através de medidas experimentais da distribuição de dose ou por simulações computacionais. Neste trabalho foram determinados os parâmetros dosimétricos da semente OncoSeed-6711 da empresa Oncura-GEHealthcare e da semente desenvolvida pelo Grupo de Dosimetria de Fontes de Braquiterapia do Centro de Tecnologia das Radiações (CTR IPEN-CNEN/SP) por simulação computacional da distribuição de dose utilizando o código MCNP5, baseado no Método de Monte Carlo. A semente 6711 foi modelada, assim como um sistema dosimétrico constituído por um objeto simulador cúbico de 30x30x30 cm3 preenchido com água. Os valores obtidos da semente 6711 foram comparados com alguns apresentados na literatura, onde o parâmetro Constante de taxa de dose apresentou erro relativo em relação ao valor publicado no TG- 43 de 0,1%, sendo que os outros parâmetros analisados também apresentaram boa concordância com os valores publicados na literatura. Deste modo, pode-se considerar que os parâmetros utilizados nas simulações (espectro, modelagem geométrica e avaliação de resultados) estão compatíveis com outros estudos, sendo estes parâmetros também utilizados nas simulações da semente do IPEN. Considerando as análises de incerteza estatística, os valores obtidos da semente do IPEN são semelhantes aos valores da semente 6711.

Tethered Monte Carlo: computing the effective potential without critical slowing down

We present Tethered Monte Carlo, a simple, general purpose method of computing the effective potential of the order parameter (Helmholtz free energy). This formalism is based on a new statistical ensemble, closely related to the micromagnetic one, but with an extended configuration space (through Creutz-like demons). Canonical averages for arbitrary values of the external magnetic field are computed without additional simulations. The method is put to work in the two-dimensional Ising model, where the existence of exact results enables us to perform high precision checks. A rather peculiar feature of our implementation, which employs a local Metropolis algorithm, is the total absence, within errors, of critical slowing down for magnetic observables. Indeed, high accuracy results are presented for lattices as large as L = 1024.