arXiv subjects

Daniel Ramos

Publications and source records attributed to Daniel Ramos.

26 records · Page 2Linked to original sources

arXiv2019

Bayesian Strategies for Likelihood Ratio Computation in Forensic Voice Comparison with Automatic Systems

This paper explores several strategies for Forensic Voice Comparison (FVC), aimed at improving the performance of the LRs when using generative Gaussian score-to-LR models. First, different anchoring strategies are proposed, with the objective of adapting the LR computation process to the case at hand, always respecting the propositions defined for the particular case. Second, a fully-Bayesian Gaussian model is used to tackle the sparsity in the training scores that is often present when the proposed anchoring strategies are used. Experiments are performed using the 2014 i-Vector challenge set-up, which presents high variability in a telephone speech context. The results show that the proposed fully-Bayesian model clearly outperforms a more common Maximum-Likelihood approach, leading to high robustness when the scores to train the model become sparse.

eess.AS↗

arXiv2019

Generative Models For Deep Learning with Very Scarce Data

The goal of this paper is to deal with a data scarcity scenario where deep learning techniques use to fail. We compare the use of two well established techniques, Restricted Boltzmann Machines and Variational Auto-encoders, as generative models in order to increase the training set in a classification framework. Essentially, we rely on Markov Chain Monte Carlo (MCMC) algorithms for generating new samples. We show that generalization can be improved comparing this methodology to other state-of-the-art techniques, e.g. semi-supervised learning with ladder networks. Furthermore, we show that RBM is better than VAE generating new samples for training a classifier with good generalization capabilities.

cs.LG↗

arXiv2018

Improving Automated Latent Fingerprint Identification using Extended Minutia Types

Latent fingerprints are usually processed with Automated Fingerprint Identification Systems (AFIS) by law enforcement agencies to narrow down possible suspects from a criminal database. AFIS do not commonly use all discriminatory features available in fingerprints but typically use only some types of features automatically extracted by a feature extraction algorithm. In this work, we explore ways to improve rank identification accuracies of AFIS when only a partial latent fingerprint is available. Towards solving this challenge, we propose a method that exploits extended fingerprint features (unusual/rare minutiae) not commonly considered in AFIS. This new method can be combined with any existing minutiae-based matcher. We first compute a similarity score based on least squares between latent and tenprint minutiae points, with rare minutiae features as reference points. Then the similarity score of the reference minutiae-based matcher at hand is modified based on a fitting error from the least square similarity stage. We use a realistic forensic fingerprint casework database in our experiments which contains rare minutiae features obtained from Guardia Civil, the Spanish law enforcement agency. Experiments are conducted using three minutiae-based matchers as a reference, namely: NIST-Bozorth3, VeriFinger-SDK and MCC-SDK. We report significant improvements in the rank identification accuracies when these minutiae matchers are augmented with our proposed algorithm based on rare minutiae features.

cs.CV↗

arXiv2017

Ricci flow on cone surfaces and a three-dimensional expanding soliton

The main objective of this thesis is the study of the evolution under the Ricci flow of surfaces with singularities of cone type. A second objective, emerged from the techniques we use, is the study of families of Ricci flow solitons in dimension 2 and 3. The Ricci flow is an evolution equation for Riemannian manifolds, introduced by R. Hamilton in 1982. It is from the achievements made by G. Perelman with this technique in 2002 when the Ricci flow has been established in a discipline itself, generating a great interest in the community. This thesis contains four original results. First result is a complete classification of solitons in smooth and cone surfaces. This classification completes the preceding results found by Hamilton, Chow and Wu and others, and we obtain explicit descriptions of all solitons in dimension 2. Second result is a Geometrization of cone surfaces by Ricci flow. This result, which uses the aforementioned first result, extends the theory of Hamilton to the singular case. This is the most comprehensive result in the thesis, for which we use and develop analysis and PDE techniques, as well as comparison geometry techniques. Third result is the existence of a Ricci flow that removes cone singularities. This clearly exposes the non-uniqueness of solutions to the flow , in analogy to the Ricci flow with cusps of P. Topping. The fourth result is the construction of a new expanding gradient Ricci soliton in dimension 3. Just as we do with solitons on cone surfaces, we give an explicit construction using techniques of phase portraits. We also prove that this is the only soliton with its topology and its lower bound of the curvature, and besides this is a critical case amongst all expanding solitons in dimension 3 with curvature bounded below.

math.DG↗

arXiv2013

Nanomechanical resonant structures in single-crystal diamond

With its host of outstanding material properties, single-crystal diamond is an attractive material for nanomechanical systems. Here, the mechanical resonance characteristics of freestanding, single-crystal diamond nanobeams fabricated by an angled-etching methodology are reported. Resonance frequencies displayed evidence of significant compressive stress in doubly clamped diamond nanobeams, while cantilever resonance modes followed the expected inverse-length-squared trend. Q-factors on the order of 104 were recorded in high vacuum. Results presented here represent initial groundwork for future diamond-based nanomechanical systems which may be applied in both classical and quantum applications.

cond-mat.mes-hall↗

arXiv2013

Gradient Ricci solitons on surfaces

We classify and expose all the gradient Ricci solitons on complete surfaces, open or closed, with curvature bounded below, and possibly with a discrete set of cone-like singular points that arise naturally. We give a precise qualitative description of each metric in terms of a phase portrait, that is the most accurate description for all cases that do not admit an explicit expression in terms of elementary functions. Our classification contains examples of smooth and conic solitons that were not described in the classic literature. We add some visual embeddings in R^3 for aesthetics.

math.DG↗

arXiv2013

An asymptotically cusped three dimensional expanding gradient Ricci soliton

We construct an expanding gradient Ricci soliton in dimension three over the topological manifold R x T^2 (the product of a line and a torus) that aproaches asymptotically a constant curvature cusp at one end, and a flat manifold on the other end. We prove that this is the only gradient soliton with this topology, provided the curvature is negatively pinched, -1/4 < sec < 0, at the time-zero manifold (normalizing the soliton to be born at time -1).

math.DG↗

arXiv2011

Smoothening cone points with Ricci flow

We consider Ricci flow on a closed surface with cone points. The main result is: given a (nonsmooth) cone metric g_0 over a closed surface there is a smooth Ricci flow g(t) defined for (0,T], with curvature unbounded above, such that g(t) tends to g_0 as t tends to 0. This result means that Ricci flow provides a way for instantaneously smoothening cone points. We follow an argument of P. Topping modifying his reasoning for cusps of negative curvature; in that sense we can consider cusps as a limiting zero-angle cone, and we generalize to any angle between 0 and 2π.

math.DG↗