arXiv subjects

Stephen Wallace

Publications and source records attributed to Stephen Wallace.

6 recordsLinked to original sources

arXiv2021

Using Causal Analysis for Conceptual Deep Learning Explanation

Model explainability is essential for the creation of trustworthy Machine Learning models in healthcare. An ideal explanation resembles the decision-making process of a domain expert and is expressed using concepts or terminology that is meaningful to the clinicians. To provide such an explanation, we first associate the hidden units of the classifier to clinically relevant concepts. We take advantage of radiology reports accompanying the chest X-ray images to define concepts. We discover sparse associations between concepts and hidden units using a linear sparse logistic regression. To ensure that the identified units truly influence the classifier's outcome, we adopt tools from Causal Inference literature and, more specifically, mediation analysis through counterfactual interventions. Finally, we construct a low-depth decision tree to translate all the discovered concepts into a straightforward decision rule, expressed to the radiologist. We evaluated our approach on a large chest x-ray dataset, where our model produces a global explanation consistent with clinical knowledge.

cs.LG↗

arXiv2021

Explaining the Black-box Smoothly- A Counterfactual Approach

We propose a BlackBox Counterfactual Explainer, designed to explain image classification models for medical applications. Classical approaches (e.g., saliency maps) that assess feature importance do not explain "how" imaging features in important anatomical regions are relevant to the classification decision. Our framework explains the decision for a target class by gradually "exaggerating" the semantic effect of the class in a query image. We adopted a Generative Adversarial Network (GAN) to generate a progressive set of perturbations to a query image, such that the classification decision changes from its original class to its negation. We used counterfactual explanations from our framework to audit a classifier trained on a chest x-ray dataset with multiple labels. We proposed clinically-relevant quantitative metrics such as cardiothoracic ratio and the score of a healthy costophrenic recess to evaluate our explanations. We conducted a human-grounded experiment with diagnostic radiology residents to compare different styles of explanations (no explanation, saliency map, cycleGAN explanation, and our counterfactual explanation) by evaluating different aspects of explanations: (1) understandability, (2) classifier's decision justification, (3) visual quality, (d) identity preservation, and (5) overall helpfulness of an explanation to the users. Our results show that our counterfactual explanation was the only explanation method that significantly improved the users' understanding of the classifier's decision compared to the no-explanation baseline. Our metrics established a benchmark for evaluating model explanation methods in medical images. Our explanations revealed that the classifier relied on clinically relevant radiographic features for its diagnostic decisions, thus making its decision-making process more transparent to the end-user.

cs.CV↗

arXiv2011

Excited state baryon spectroscopy from lattice QCD with spin identification

Lattice QCD calculations are presented for the spectra of N* excited states with spins up to J = 7/2. Ambiguities of the standard method of spin identification are shown to be overcome by the use of lattice operators that transform according to SU(2) symmetry restricted to the lattice. Such operators are labeled by their continuum spins. Overlaps of the operators with the states obtained by diagonalizing matrices of correlation functions provide a clear link between continuum spins and lattice states, allowing spins to be identified. Evidence for an approximate realization of rotational symmetry in the N* spectrum is presented. In simulations with pion mass = 392 MeV, the low-lying excited states of lattice QCD are found to have the same quantum numbers as the states of SU(6)xO(3) symmetry. The lattice spectra are inconsistent with either a quark-diquark model or parity doubling of states and they suggest that the J = 1/2 Roper resonance may have a complex structure consisting of contributions from L=0, 1 and 2.

hep-lat↗

arXiv2009

Exploring the Spectrum of QCD using the Lattice

The calculation of the spectrum of QCD is key to an understanding of the strong interactions, and vital if we are to capitalize on the experimental study of the spectrum. In this paper, we describe progress towards understanding the spectrum of resonances of both mesons and baryons from lattice QCD, focusing in particular on the resonances of the $I=1/2$ nucleon states, and of charmonium mesons composed of the heavy charmed quarks.

hep-lat↗

arXiv2008

First results from 2+1 dynamical quark flavors on an anisotropic lattice: light-hadron spectroscopy and setting the strange-quark mass

We present the first light-hadron spectroscopy on a set of $N_f=2+1$ dynamical, anisotropic lattices. A convenient set of coordinates that parameterize the two-dimensional plane of light and strange-quark masses is introduced. These coordinates are used to extrapolate data obtained at the simulated values of the quark masses to the physical light and strange-quark point. A measurement of the Sommer scale on these ensembles is made, and the performance of the hybrid Monte Carlo algorithm used for generating the ensembles is estimated.

hep-lat↗

arXiv2005

Combining Quark and Link Smearing to Improve Extended Baryon Operators

The effects of Gaussian quark-field smearing and analytic stout-link smearing on the correlations of gauge-invariant extended baryon operators are studied. Gaussian quark-field smearing substantially reduces contributions from the short wavelength modes of the theory, while stout-link smearing significantly reduces the noise from the stochastic evaluations. The use of gauge-link smearing is shown to be crucial for baryon operators constructed of covariantly-displaced quark fields. Preferred smearing parameters are determined for a lattice spacing a_s ~ 0.1 fm.

hep-lat↗