-
Heckman Selection Contaminated Normal Model
Authors:
Heeju Lim,
Jose Alejandro Ordonez,
Victor H. Lachos,
Antonio Punzo
Abstract:
The Heckman selection model is one of the most well-renounced econometric models in the analysis of data with sample selection. This model is designed to rectify sample selection biases based on the assumption of bivariate normal error terms. However, real data diverge from this assumption in the presence of heavy tails and/or atypical observations. Recently, this assumption has been relaxed via a…
▽ More
The Heckman selection model is one of the most well-renounced econometric models in the analysis of data with sample selection. This model is designed to rectify sample selection biases based on the assumption of bivariate normal error terms. However, real data diverge from this assumption in the presence of heavy tails and/or atypical observations. Recently, this assumption has been relaxed via a more flexible Student's t-distribution, which has appealing statistical properties. This paper introduces a novel Heckman selection model using a bivariate contaminated normal distribution for the error terms. We present an efficient ECM algorithm for parameter estimation with closed-form expressions at the E-step based on truncated multinormal distribution formulas. The identifiability of the proposed model is also discussed, and its properties have been examined. Through simulation studies, we compare our proposed model with the normal and Student's t counterparts and investigate the finite-sample properties and the variation in missing rate. Results obtained from two real data analyses showcase the usefulness and effectiveness of our model. The proposed algorithms are implemented in the R package HeckmanEM.
△ Less
Submitted 18 September, 2024;
originally announced September 2024.
-
On the KG-constrained Bekenstein's disformal transformation of the Einstein-Hilbert action
Authors:
Allan L. Alinea,
Joshwa DJ. Ordonez
Abstract:
Motivated by an inclination for symmetry and possible extension of the General Theory of Relativity within the framework of Scalar Theory, we investigate the Bekenstein's disformal transformation of the Einstein-Hilbert action. Owing to the complicated combinations of second order metric derivatives encoded in the Ricci scalar of the action, such a transformation yields an unwieldy expression. To…
▽ More
Motivated by an inclination for symmetry and possible extension of the General Theory of Relativity within the framework of Scalar Theory, we investigate the Bekenstein's disformal transformation of the Einstein-Hilbert action. Owing to the complicated combinations of second order metric derivatives encoded in the Ricci scalar of the action, such a transformation yields an unwieldy expression. To ``tame'' the transformed action, we exploit the Klein-Gordon (KG) conformal-disformal constraint previously discovered in the study of the invariance of the massless Klein-Gordon equation under disformal transformation. The result upon its application is a surprisingly much more concise and simple action in four spacetime dimensions containing three out of four sub-Lagrangians in the Horndeski action, and three beyond-Horndeski terms. The latter group of terms may be attributed to the kinetic-{term} dependence of the conformal and disformal factors in the Bekenstein's disformal transformation. Going down to three dimensions, we find a relatively simpler resulting action but the signature of the three ``extraneous'' terms remains. Lastly, in two dimensions, we find an invariant action consistent with its topological nature in these dimensions.
△ Less
Submitted 4 March, 2025; v1 submitted 13 August, 2024;
originally announced August 2024.
-
On the KG-constrained general disformal transformation of the Einstein-Hilbert action
Authors:
Allan L. Alinea,
Joshwa DJ. Ordonez
Abstract:
Although with great successes in explaining phenomena and natural behaviour involving the Universe or a part thereof, the General Theory of Relativity is far from a complete theory. Focusing on its extension within the framework of scalar tensor theory, we investigate the transformation of the Einstein-Hilbert action under the general disformal transformation coupled with a set of triple KG constr…
▽ More
Although with great successes in explaining phenomena and natural behaviour involving the Universe or a part thereof, the General Theory of Relativity is far from a complete theory. Focusing on its extension within the framework of scalar tensor theory, we investigate the transformation of the Einstein-Hilbert action under the general disformal transformation coupled with a set of triple KG constraints. The idea is that the general disformal transformation leads to an extremely complicated action necessitating a set of constraints to tame the resulting form. Motivated by previous studies on the invertibility and invariance of the massless Klein-Gordon equation under the general disformal transformation, we identify three constraints that significantly simplifies the transformed Einstein-Hilbert action. In four spacetime dimensions, we find that the transformed action is a sum of the original action and a disformal contribution involving a six-term Lagrangian that includes the Ricci tensor coupled to a sum of derivatives of scalar fields and kinetic terms. In three spacetime dimensions, the disformal contribution becomes a five-term Lagrangian. Lastly, in two spacetime dimensions, the Einstein-Hilbert action is invariant under the constrained disformal transformation.
△ Less
Submitted 22 September, 2024; v1 submitted 12 August, 2024;
originally announced August 2024.
-
Detect, Retrieve, Comprehend: A Flexible Framework for Zero-Shot Document-Level Question Answering
Authors:
Tavish McDonald,
Brian Tsan,
Amar Saini,
Juanita Ordonez,
Luis Gutierrez,
Phan Nguyen,
Blake Mason,
Brenda Ng
Abstract:
Researchers produce thousands of scholarly documents containing valuable technical knowledge. The community faces the laborious task of reading these documents to identify, extract, and synthesize information. To automate information gathering, document-level question answering (QA) offers a flexible framework where human-posed questions can be adapted to extract diverse knowledge. Finetuning QA s…
▽ More
Researchers produce thousands of scholarly documents containing valuable technical knowledge. The community faces the laborious task of reading these documents to identify, extract, and synthesize information. To automate information gathering, document-level question answering (QA) offers a flexible framework where human-posed questions can be adapted to extract diverse knowledge. Finetuning QA systems requires access to labeled data (tuples of context, question and answer). However, data curation for document QA is uniquely challenging because the context (i.e. answer evidence passage) needs to be retrieved from potentially long, ill-formatted documents. Existing QA datasets sidestep this challenge by providing short, well-defined contexts that are unrealistic in real-world applications. We present a three-stage document QA approach: (1) text extraction from PDF; (2) evidence retrieval from extracted texts to form well-posed contexts; (3) QA to extract knowledge from contexts to return high-quality answers -- extractive, abstractive, or Boolean. Using QASPER for evaluation, our detect-retrieve-comprehend (DRC) system achieves a +7.19 improvement in Answer-F1 over existing baselines while delivering superior context selection. Our results demonstrate that DRC holds tremendous promise as a flexible framework for practical scientific document QA.
△ Less
Submitted 11 December, 2023; v1 submitted 4 October, 2022;
originally announced October 2022.
-
Spatial Censored Regression Models in R: The CensSpatial package
Authors:
Jose A. Ordonez,
Christian E. Galarza,
Victor H. Lachos
Abstract:
CensSpatial is an R package for analyzing spatial censored data through linear models. It offers a set of tools for simulating, estimating, making predictions, and performing local influence diagnostics for outlier detection. The package provides four algorithms for estimation and prediction. One of them is based on the stochastic approximation of the EM (SAEM) algorithm, which allows easy and fas…
▽ More
CensSpatial is an R package for analyzing spatial censored data through linear models. It offers a set of tools for simulating, estimating, making predictions, and performing local influence diagnostics for outlier detection. The package provides four algorithms for estimation and prediction. One of them is based on the stochastic approximation of the EM (SAEM) algorithm, which allows easy and fast estimation of the parameters of linear spatial models when censoring is present. The package provides worthy measures to perform diagnostic analysis using the Hessian matrix of the completed log-likelihood function. This work is divided into two parts. The first part discusses and illustrates the utilities that the package offers for estimating and predicting spatial censored data. The second one describes the valuable tools to perform diagnostic analysis. Several examples in spatial environmental data are also provided.
△ Less
Submitted 11 October, 2021;
originally announced October 2021.
-
Credit Assignment in Neural Networks through Deep Feedback Control
Authors:
Alexander Meulemans,
Matilde Tristany Farinha,
Javier García Ordóñez,
Pau Vilimelis Aceituno,
João Sacramento,
Benjamin F. Grewe
Abstract:
The success of deep learning sparked interest in whether the brain learns by using similar techniques for assigning credit to each synaptic weight for its contribution to the network output. However, the majority of current attempts at biologically-plausible learning methods are either non-local in time, require highly specific connectivity motives, or have no clear link to any known mathematical…
▽ More
The success of deep learning sparked interest in whether the brain learns by using similar techniques for assigning credit to each synaptic weight for its contribution to the network output. However, the majority of current attempts at biologically-plausible learning methods are either non-local in time, require highly specific connectivity motives, or have no clear link to any known mathematical optimization method. Here, we introduce Deep Feedback Control (DFC), a new learning method that uses a feedback controller to drive a deep neural network to match a desired output target and whose control signal can be used for credit assignment. The resulting learning rule is fully local in space and time and approximates Gauss-Newton optimization for a wide range of feedback connectivity patterns. To further underline its biological plausibility, we relate DFC to a multi-compartment model of cortical pyramidal neurons with a local voltage-dependent synaptic plasticity rule, consistent with recent theories of dendritic processing. By combining dynamical system theory with mathematical optimization theory, we provide a strong theoretical foundation for DFC that we corroborate with detailed results on toy experiments and standard computer-vision benchmarks.
△ Less
Submitted 17 January, 2022; v1 submitted 15 June, 2021;
originally announced June 2021.
-
Suboptimal multirate MPC for five-level inverters
Authors:
Joaquin G. Ordonez,
Francisco Gordillo,
Pablo Montero-Robina,
Daniel Limon
Abstract:
The application of multilevel converters to renewable energy systems is a growing topic due to their advantages in energy efficiency. Regarding its control, model predictive control (MPC) has become very appealing due to its natural consideration of discrete inputs, its optimization capability, and the present-day availability of powerful processing hardware. The main drawback of MPC compared to o…
▽ More
The application of multilevel converters to renewable energy systems is a growing topic due to their advantages in energy efficiency. Regarding its control, model predictive control (MPC) has become very appealing due to its natural consideration of discrete inputs, its optimization capability, and the present-day availability of powerful processing hardware. The main drawback of MPC compared to other control techniques in this field is that the control input is held constant during the sampling period, and it is usually difficult or even impossible to reduce this sampling period because of hardware limitations. For this reason, a multirate MPC algorithm is proposed, which allows to change the control input several times within the sampling period. The optimization problem is simplified and made suboptimal to substantially decrease computational burden. This approach is tested in simulation on a three-phase, five-level diode-clamped converter (DCC) operating in inverted mode with a three-phase resistive load. Results show significant reduction in harmonic distortion at the cost of an increase in the number of commutations with respect to a standard MPC operating at the same sampling period.
△ Less
Submitted 5 April, 2021;
originally announced April 2021.
-
Grid Search Hyperparameter Benchmarking of BERT, ALBERT, and LongFormer on DuoRC
Authors:
Alex John Quijano,
Sam Nguyen,
Juanita Ordonez
Abstract:
The purpose of this project is to evaluate three language models named BERT, ALBERT, and LongFormer on the Question Answering dataset called DuoRC. The language model task has two inputs, a question, and a context. The context is a paragraph or an entire document while the output is the answer based on the context. The goal is to perform grid search hyperparameter fine-tuning using DuoRC. Pretrain…
▽ More
The purpose of this project is to evaluate three language models named BERT, ALBERT, and LongFormer on the Question Answering dataset called DuoRC. The language model task has two inputs, a question, and a context. The context is a paragraph or an entire document while the output is the answer based on the context. The goal is to perform grid search hyperparameter fine-tuning using DuoRC. Pretrained weights of the models are taken from the Huggingface library. Different sets of hyperparameters are used to fine-tune the models using two versions of DuoRC which are the SelfRC and the ParaphraseRC. The results show that the ALBERT (pretrained using the SQuAD1 dataset) has an F1 score of 76.4 and an accuracy score of 68.52 after fine-tuning on the SelfRC dataset. The Longformer model (pretrained using the SQuAD and SelfRC datasets) has an F1 score of 52.58 and an accuracy score of 46.60 after fine-tuning on the ParaphraseRC dataset. The current results outperformed the results from the previous model by DuoRC.
△ Less
Submitted 29 March, 2021; v1 submitted 15 January, 2021;
originally announced January 2021.
-
Objective Bayesian analysis for spatial Student-t regression models
Authors:
Jose A. Ordoñez,
Marcos O. Prates,
Larissa A. Matos,
Victor H. Lachos
Abstract:
The choice of the prior distribution is a key aspect of Bayesian analysis. For the spatial regression setting a subjective prior choice for the parameters may not be trivial, from this perspective, using the objective Bayesian analysis framework a reference is introduced for the spatial Student-t regression model with unknown degrees of freedom. The spatial Student-t regression model poses two mai…
▽ More
The choice of the prior distribution is a key aspect of Bayesian analysis. For the spatial regression setting a subjective prior choice for the parameters may not be trivial, from this perspective, using the objective Bayesian analysis framework a reference is introduced for the spatial Student-t regression model with unknown degrees of freedom. The spatial Student-t regression model poses two main challenges when eliciting priors: one for the spatial dependence parameter and the other one for the degrees of freedom. It is well-known that the propriety of the posterior distribution over objective priors is not always guaranteed, whereas the use of proper prior distributions may dominate and bias the posterior analysis. In this paper, we show the conditions under which our proposed reference prior yield to a proper posterior distribution. Simulation studies are used in order to evaluate the performance of the reference prior to a commonly used vague proper prior.
△ Less
Submitted 8 April, 2020;
originally announced April 2020.
-
Experimental realisation of tunable ferroelectric/superconductor (BTO/YBCO)N/STO 1D photonic crystals in the whole visible spectrum
Authors:
Luz E. González,
John E. Ordoñez,
Carlos A. Melo-Luna,
Evelyn Mendoza,
David Reyes,
Gustavo Zambrano,
Nelson Porras-Montenegro,
Juan C. Granada,
Maria E. Gómez,
John H. Reina
Abstract:
Emergent technologies that make use of novel materials and quantum properties of light states are at the forefront in the race for the physical implementation, encoding and transmission of information. Photonic crystals (PCs) enter this paradigm with optical materials that allow the control of light propagation and can be used for optical communication, and photonics and electronics integration ma…
▽ More
Emergent technologies that make use of novel materials and quantum properties of light states are at the forefront in the race for the physical implementation, encoding and transmission of information. Photonic crystals (PCs) enter this paradigm with optical materials that allow the control of light propagation and can be used for optical communication, and photonics and electronics integration making use of materials ranging from semiconductors, to metals, metamaterials, and topological insulators, to mention but a few. In particular, here we show how designer superconductor materials integrated into PCs fabrication allow for an extraordinary reduction of electromagnetic waves damping and possibilitate their optimal propagation and tuning through the structure, below critical superconductor temperature. We experimentally demonstrate, for the first time, a successful integration of ferroelectric and superconductor materials into a one-dimensional (1D) PC composed of (BTO/YBCO)N/STO bilayers that work in the whole visible spectrum, and below (and above) critical superconductor temperature (measured in the 10-300 K range). Theoretical calculations support, for different number of bilayers N, the effectiveness of the produced 1D PCs and pave the way for novel optoelectronics integration and information processing in the visible spectrum at low temperature, while preserving their electric and optical properties.
△ Less
Submitted 23 September, 2019;
originally announced September 2019.
-
Deep Hubble Space Telescope Imaging of Globular Clusters Towards the Galactic Bulge: Observations, Data Reduction and Color-Magnitude Diagrams
Authors:
Roger E. Cohen,
Francesco Mauro,
Javier Alonso-García,
Maren Hempel,
Ata Sarajedini,
Antonio J. Ordoñez,
Douglas Geisler,
Jason S. Kalirai
Abstract:
The Galactic globular clusters (GGCs) located towards the Galactic bulge have generally been excluded from large-scale photometric GGC surveys due to severe total and differential extinction. Here, we present an overview of a Hubble Space Telescope program designed to obtain deep, high spatial resolution multiband imaging of 16 poorly studied GGCs located towards the inner Galactic bulge and disk.…
▽ More
The Galactic globular clusters (GGCs) located towards the Galactic bulge have generally been excluded from large-scale photometric GGC surveys due to severe total and differential extinction. Here, we present an overview of a Hubble Space Telescope program designed to obtain deep, high spatial resolution multiband imaging of 16 poorly studied GGCs located towards the inner Galactic bulge and disk. In this first paper of a series resulting from these observations, we give an overview of target cluster selection, observations and data reduction procedures for optimizing the resulting photometric catalogs. Artificial star tests are used to compare the respective advantages of different data reduction strategies in terms of photometric and astrometric precision and photometric incompleteness. We present the resulting color-magnitude diagrams (CMDs) of all target clusters in several color-magnitude planes, along with CMDs of comparison fields from parallel observations. For each target cluster, we summarize existing studies, and discuss their CMDs qualitatively in the context of these results.
△ Less
Submitted 28 May, 2018;
originally announced May 2018.
-
The pulsating variable star population in DDO210
Authors:
Antonio J. Ordoñez,
Ata Sarajedini
Abstract:
We have probed the pulsating variable star content of the isolated Local Group dwarf galaxy, DDO210 (Aquarius), using archival Advanced Camera for Surveys/$Hubble$ $Space$ $Telescope$ imaging in the F475W and F814W passbands. We find a total of 32 RR Lyrae stars (24 ab-type, 8 c-type) and 75 Cepheid variables. The mean periods of the ab-type and c-type RR Lyrae stars are calculated to be…
▽ More
We have probed the pulsating variable star content of the isolated Local Group dwarf galaxy, DDO210 (Aquarius), using archival Advanced Camera for Surveys/$Hubble$ $Space$ $Telescope$ imaging in the F475W and F814W passbands. We find a total of 32 RR Lyrae stars (24 ab-type, 8 c-type) and 75 Cepheid variables. The mean periods of the ab-type and c-type RR Lyrae stars are calculated to be $\langle$P$_{\mathrm{ab}}\rangle = 0.609\pm0.011$ and $\langle$P$_{\mathrm{c}}\rangle = 0.359\pm0.025$ days, respectively. The light curve properties of the fundamental mode RR Lyrae stars yield a mean metallicity of $\langle$[Fe/H]$\rangle$ = -1.63$\pm$0.11 dex for this ancient population, consistent with a recent synthetic colour-magnitude diagram analysis. We find this galaxy to be Oosterhoff-intermediate and lacking in high-amplitude, short-period ab-type RR Lyrae, consistent with behavior recently observed for many dwarf spheroidals and ultra-faint dwarfs in the Local Group. We find a distance modulus of $μ= 25.07\pm 0.12$ as determined by the RR Lyrae stars, slightly larger but agreeing with recent distance estimates from the red giant branch tip. We also find a sizable population of Cepheid variables in this galaxy. We provide evidence in favor of most if not all of these stars being short-period classical Cepheids. Assuming all of these stars to be classical Cepheids, we find that most of these Cepheids are $\sim$300 Myr old, with the youngest Cepheids being offset from the older Cepheids and the centre of the galaxy. We conclude that this may have resulted from a migration of star formation in DDO210.
△ Less
Submitted 23 October, 2015;
originally announced October 2015.
-
Population Effects on the Metallicity Distribution Function Derived From the Red Giant Branch
Authors:
Antonio J. Ordoñez,
Ata Sarajedini
Abstract:
We have tested the reliability of the red giant branch (RGB) as a metallicity indicator accounting for observational errors as well as the complexity of star formation histories (SFHs) and chemical evolution histories observed in various stellar systems. We generate model color-magnitude diagrams (CMDs) produced with a variety of evolutionary histories and compare the resultant metallicity estimat…
▽ More
We have tested the reliability of the red giant branch (RGB) as a metallicity indicator accounting for observational errors as well as the complexity of star formation histories (SFHs) and chemical evolution histories observed in various stellar systems. We generate model color-magnitude diagrams (CMDs) produced with a variety of evolutionary histories and compare the resultant metallicity estimates from the colors and magnitudes of RGB stars to the true input metallicities. We include realistic models for photometric errors and completeness in our synthetic CMDs. As expected, for simple simple stellar populations dominated by old stars, the RGB provides a very accurate estimate of the modular metallicity value for a population. An error in the age of a system targeted for this type of study may produce metallicity errors of a few tenths of a dex. The size of this metallicity error depends linearly on the age error, and we find this dependence to be stronger with more precise photometry. If the population has experienced any significant star formation within the last $\sim$6 Gyr, the metallicity estimates, [M/H], derived from the RGB may be in error by up to $\sim$0.5 dex. Perhaps the most important consideration for this technique is an accurate, independent estimate of the average age for the target stellar system, especially if it is probable that a significant fraction of the population formed less than $\sim$6 Gyr ago.
△ Less
Submitted 6 April, 2015; v1 submitted 3 April, 2015;
originally announced April 2015.
-
The RR Lyrae Variable Population in the Phoenix Dwarf Galaxy
Authors:
Antonio J. Ordoñez,
Soung-Chul Yang,
Ata Sarajedini
Abstract:
We present the first detailed study of the RR Lyrae variable population in the Local Group dSph/dIrr transition galaxy, Phoenix, using previously obtained HST/WFPC2 observations of the galaxy. We utilize template light curve fitting routines to obtain best fit light curves for RR Lyrae variables in Phoenix. Our technique has identified 78 highly probable RR Lyrae stars (54 ab-type; 24 c-type) with…
▽ More
We present the first detailed study of the RR Lyrae variable population in the Local Group dSph/dIrr transition galaxy, Phoenix, using previously obtained HST/WFPC2 observations of the galaxy. We utilize template light curve fitting routines to obtain best fit light curves for RR Lyrae variables in Phoenix. Our technique has identified 78 highly probable RR Lyrae stars (54 ab-type; 24 c-type) with about 40 additional candidates. We find mean periods for the two populations of $\langle P_{ab}\rangle = 0.60 \pm 0.03$ days and $\langle P_{c}\rangle = 0.353 \pm 0.002$ days. We use the properties of these light curves to extract, among other things, a metallicity distribution function for ab-type RR Lyrae. Our analysis yields a mean metallicity of $\langle [Fe/H]\rangle = -1.68 \pm 0.06$ dex for the RRab stars. From the mean period and metallicity calculated from the ab-type RR Lyrae, we conclude that Phoenix is more likely of intermediate Oosterhoff type; however the morphology of the Bailey diagram for Phoenix RR Lyraes appears similar to that of an Oosterhoff type I system. Using the RRab stars, we also study the chemical enrichment law for Phoenix. We find that our metallicity distribution is reasonably well fitted by a closed-box model. The parameters of this model are compatible with the findings of Hidalgo et al. (2009) further supporting the idea that Phoenix appears to have been chemically enriched as a closed-box-like system during the early stage of its formation and evolution.
△ Less
Submitted 3 April, 2014;
originally announced April 2014.
