arXiv subjects

Javier Gonzalez

Publications and source records attributed to Javier Gonzalez.

29 records · Page 2Linked to original sources

arXiv2020

Polarization calibration techniques for new-generation VLBI

The calibration and analysis of polarization observations in Very Long Baseline Interferometry (VLBI) requires the use of specific algorithms that suffer from several limitations, closely related to assumptions in the data properties that may not hold in observations taken with new-generation VLBI equipment. Nowadays, the instantaneous bandwidth achievable with VLBI backends can be as high as several GHz, covering several radio bands simultaneously. In addition, the sensitivity of VLBI observations with state-of-the-art equipment may reach dynamic ranges of tens of thousands, both in total intensity and in polarization. In this paper, we discuss the impact of the limitations of common VLBI polarimetry algorithms on narrow-field observations taken with modern VLBI arrays (from the VLBI Global Observing System, VGOS, to the Event Horizon Telescope, EHT) and present new software that overcomes these limitations. In particular, our software is able to perform a simultaneous fit of multiple calibrator sources, include non-linear terms in the model of the instrumental polarization and use a self-calibration approach for the estimate of the polarization leakage in the antenna receivers.

astro-ph.IM↗

arXiv2020

BOSS: Bayesian Optimization over String Spaces

This article develops a Bayesian optimization (BO) method which acts directly over raw strings, proposing the first uses of string kernels and genetic algorithms within BO loops. Recent applications of BO over strings have been hindered by the need to map inputs into a smooth and unconstrained latent space. Learning this projection is computationally and data-intensive. Our approach instead builds a powerful Gaussian process surrogate model based on string kernels, naturally supporting variable length inputs, and performs efficient acquisition function maximization for spaces with syntactical constraints. Experiments demonstrate considerably improved optimization over existing approaches across a broad range of constraints, including the popular setting where syntax is governed by a context-free grammar.

cs.LG↗

arXiv2020

Automatic Discovery of Privacy-Utility Pareto Fronts

Differential privacy is a mathematical framework for privacy-preserving data analysis. Changing the hyperparameters of a differentially private algorithm allows one to trade off privacy and utility in a principled way. Quantifying this trade-off in advance is essential to decision-makers tasked with deciding how much privacy can be provided in a particular application while maintaining acceptable utility. Analytical utility guarantees offer a rigorous tool to reason about this trade-off, but are generally only available for relatively simple problems. For more complex tasks, such as training neural networks under differential privacy, the utility achieved by a given algorithm can only be measured empirically. This paper presents a Bayesian optimization methodology for efficiently characterizing the privacy--utility trade-off of any differentially private algorithm using only empirical measurements of its utility. The versatility of our method is illustrated on a number of machine learning tasks involving multiple models, optimizers, and datasets.

stat.ML↗

arXiv2020

BINOCULARS for Efficient, Nonmyopic Sequential Experimental Design

Finite-horizon sequential experimental design (SED) arises naturally in many contexts, including hyperparameter tuning in machine learning among more traditional settings. Computing the optimal policy for such problems requires solving Bellman equations, which are generally intractable. Most existing work resorts to severely myopic approximations by limiting the decision horizon to only a single time-step, which can underweight exploration in favor of exploitation. We present BINOCULARS: Batch-Informed NOnmyopic Choices, Using Long-horizons for Adaptive, Rapid SED, a general framework for deriving efficient, nonmyopic approximations to the optimal experimental policy. Our key idea is simple and surprisingly effective: we first compute a one-step optimal batch of experiments, then select a single point from this batch to evaluate. We realize BINOCULARS for Bayesian optimization and Bayesian quadrature -- two notable SED problems with radically different objectives -- and demonstrate that BINOCULARS significantly outperforms myopic alternatives in real-world scenarios.

cs.LG↗

arXiv2019

Meta-Surrogate Benchmarking for Hyperparameter Optimization

Despite the recent progress in hyperparameter optimization (HPO), available benchmarks that resemble real-world scenarios consist of a few and very large problem instances that are expensive to solve. This blocks researchers and practitioners not only from systematically running large-scale comparisons that are needed to draw statistically significant results but also from reproducing experiments that were conducted before. This work proposes a method to alleviate these issues by means of a meta-surrogate model for HPO tasks trained on off-line generated data. The model combines a probabilistic encoder with a multi-task model such that it can generate inexpensive and realistic tasks of the class of problems of interest. We demonstrate that benchmarking HPO methods on samples of the generative model allows us to draw more coherent and statistically significant conclusions that can be reached orders of magnitude faster than using the original tasks. We provide evidence of our findings for various HPO methods on a wide class of problems.

cs.LG↗

arXiv2019

Results on Proton-Irradiated 3D Pixel Sensors Interconnected to RD53A Readout ASIC

Test beam results obtained with 3D pixel sensors bump-bonded to the RD53A prototype readout ASIC are reported. Sensors from FBK (Italy) and IMB-CNM (Spain) have been tested before and after proton-irradiation to an equivalent fluence of about $1$ $\times$ $10^{16}$ $\text{n}_{\text{eq}}$ cm$^{-2}$ (1 MeV equivalent neutrons). This is the first time that one single collecting electrode fine pitch 3D sensors are irradiated up to such fluence bump-bonded to a fine pitch ASIC. The preliminary analysis of the collected data shows no degradation on the hit detection efficiencies of the tested sensors after high energy proton irradiation, demonstrating the excellent radiation tolerance of the 3D pixel sensors. Thus, they will be excellent candidates for the extreme radiation environment at the innermost layers of the HL-LHC experiments.

physics.ins-det↗

arXiv2019

A Three-dimensional Reconstruction of Cosmic Ray Events in IceCube

The IceCube Neutrino Observatory at the geographic South Pole consists of two components, a km2 surface array IceTop and a km3 in-ice array between 1.5 and 2.5 km below the surface. Cosmic ray events with primary energy above a few tens of TeV may trigger both the IceTop and in-ice array and leave a three-dimensional footprint of the electromagnetic and muonic components in the extensive air shower. A new reconstruction based on the minimization of a unified likelihood function involving quantities measured by both IceTop and in-ice detectors was developed. This report describes the new reconstruction algorithm and summarizes its performance tested with Monte Carlo events under two different containment conditions. The advantages of the new reconstruction are discussed in comparison with reconstructions that use IceTop or in-ice data separately. Some possible improvements are also summarized.

astro-ph.HE↗

arXiv2017

Preferential Bayesian Optimization

Bayesian optimization (BO) has emerged during the last few years as an effective approach to optimizing black-box functions where direct queries of the objective are expensive. In this paper we consider the case where direct access to the function is not possible, but information about user preferences is. Such scenarios arise in problems where human preferences are modeled, such as A/B tests or recommender systems. We present a new framework for this scenario that we call Preferential Bayesian Optimization (PBO) which allows us to find the optimum of a latent function that can only be queried through pairwise comparisons, the so-called duels. PBO extends the applicability of standard BO ideas and generalizes previous discrete dueling approaches by modeling the probability of the winner of each duel by means of a Gaussian process model with a Bernoulli likelihood. The latent preference function is used to define a family of acquisition functions that extend usual policies used in BO. We illustrate the benefits of PBO in a variety of experiments, showing that PBO needs drastically fewer comparisons for finding the optimum. According to our experiments, the way of modeling correlations in PBO is key in obtaining this advantage.

stat.ML↗

arXiv2013

Capacity Analysis of IEEE 802.11ah WLANs for M2M Communications

Focusing on the increasing market of the sensors and actuators networks, the IEEE 802.11ah Task Group is currently working on the standardization of a new amendment. This new amendment will operate at the sub-1GHz band, ensure transmission ranges up to 1 Km, data rates above 100 kbps and very low power operation. With IEEE 802.11ah, the WLANs will offer a solution for applications such as smart metering, plan automation, eHealth or surveillance. Moreover, thanks to a hierarchical signalling, the IEEE 802.11ah will be able to manage a higher number of stations (STAs) and improve the 802.11 Power Saving Mechanisms. In order to support a high number of STAs, two different signalling modes are proposed, TIM and Non-TIM Offset. In this paper we present a theoretical model to predict the maximum number of STAs supported by both modes depending on the traffic load and the data rate used. Moreover, the IEEE 802.11ah performance and energy consumption for both signalling modes and for different traffic patterns and data rates is evaluated. Results show that both modes achieve similar Packet Delivery Ratio values but the energy consumed with the TIM Offset is, in average, a 11.7% lower.

cs.NI↗

arXiv2009

Alignment of Silicon tracking systems R&D on Semitransparent Microstrip Sensors

We summarize the R&D activities on a novel semitransparent microstrip sensor to be used on laser-based alignment systems for silicon trackers. The new sensor is used both for particle tracking and laser detection. The aim of this research line is to increase the optical transmittance (T) of Silicon microstrips detectors to infrared light, introducing minor modifications to the sensor design still suitable for its industrial production. The optical simulations used in the sensor design have been experimentally validated against several patterned material samples. This activities have been carried out in the context of SiLC collaboration for the next International Linear Collider.

physics.ins-det↗

arXiv2007

Applications of the Characteristic Theory to the Madelung-de Broglie-Bohm System of Partial Differential Equations: The Guiding Equation as the Characteristic Velocity

First, we use the theory of characteristics of first order partial differential equations to derive the guiding equation directly from the Quantum Evolution Equation (QEE). After obtaining the general result, we apply it to a set of evolution equations (Schroedinger, Pauli, Klein-Gordon, Dirac) to show how the guiding equation is, actually, the characteristic velocity of the corresponding matter field equations.

quant-ph↗