Source author record

Lorenzo Mannelli

Lorenzo Mannelli appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

hep-ph hep-th hep-lat Computer Vision eess.IV Machine Learning

Catalog footprint

What is connected

7works

6topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2021arXiv

Detecting Spurious Correlations with Sanity Tests for Artificial Intelligence Guided Radiology Systems

Artificial intelligence (AI) has been successful at solving numerous problems in machine perception. In radiology, AI systems are rapidly evolving and show progress in guiding treatment decisions, diagnosing, localizing disease on medical images, and improving radiologists' efficiency. A critical component to deploying AI in radiology is to gain confidence in a developed system's efficacy and safety. The current gold standard approach is to conduct an analytical validation of performance on a generalization dataset from one or more institutions, followed by a clinical validation study of the system's efficacy during deployment. Clinical validation studies are time-consuming, and best practices dictate limited re-use of analytical validation data, so it is ideal to know ahead of time if a system is likely to fail analytical or clinical validation. In this paper, we describe a series of sanity tests to identify when a system performs well on development data for the wrong reasons. We illustrate the sanity tests' value by designing a deep learning system to classify pancreatic cancer seen in computed tomography scans.

preprint2012arXiv

Three-Prong Distribution of Massive Narrow QCD Jets

We study the planar-flow distributions of narrow, highly boosted, massive QCD jets. Using the factorization properties of QCD in the collinear limit, we compute the planar-flow jet function from the one-to-three splitting function at tree-level. We derive the leading-log behavior of the jet function analytically. We also compare our semi-analytic jet function with parton-shower predictions using various generators.

preprint2010arXiv

Covariant Description of Flavor Conversion in the LHC Era

A simple covariant formalism to describe flavor and CP violation in the left-handed quark sector in a model independent way is provided. The introduction of a covariant basis, which makes the standard model approximate symmetry structure manifest, leads to a physical and transparent picture of flavor conversion processes. Our method is particularly useful to derive robust bounds on models with arbitrary mechanisms of alignment. Known constraints on flavor violation in the K and D systems are reproduced in a straightforward manner. Assumptions-free limits, based on top flavor violation at the LHC, are then obtained. In the absence of signal, with 100 fb^{-1} of data, the LHC will exclude weakly coupled (strongly coupled) new physics up to a scale of 0.6 TeV (7.6 TeV), while at present no general constraint can be set related to Delta t=1 processes. LHC data will constrain Delta F=2 contributions via same-sign tops signal, with a model independent exclusion region of 0.08 TeV (1.0 TeV). However, in this case, stronger bounds are found from the study of CP violation in D-bar D mixing with a scale of 0.57 TeV (7.2 TeV). In addition, we apply our analysis to models of supersymmetry and warped extra dimension. The minimal flavor violation framework is also discussed, where the formalism allows to distinguish between the linear and generic non-linear limits within this class of models.

preprint2010arXiv

Covariant Description of Flavor Violation at the LHC

A simple formalism to describe flavor and CP violation in a model independent way is provided. Our method is particularly useful to derive robust bounds on models with arbitrary mechanisms of alignment. Known constraints on flavor violation in the K and D systems are reproduced in a straightforward and covariant manner. Assumptions-free limits, based on top flavor violation at the LHC, are then obtained. In the absence of signal, with 100 fb^{-1} of data, the LHC will exclude weakly coupled (strongly coupled) new physics up to a scale of 0.6 TeV (7.6 TeV), while at present no general constraint can be set related to Delta t=1 processes. Delta F=2 contributions will be constrained via same-sign tops signal, with a model independent exclusion region of 0.08 TeV (1.0 TeV). However, in this case, stronger bounds are found from the study of CP violation in D-bar D mixing with a scale of 0.57 TeV (7.2 TeV). We also apply our analysis to supersymmetric and warped extra dimension models.

preprint2009arXiv

Four-dimensional N=1 super Yang-Mills from matrix model

We consider a supersymmetric matrix quantum mechanics. This is obtained by adding Myers and mass terms to the dimensional reduction of 4d N=1 super Yang-Mills theory to one dimension. Using this model we construct 4d N=1 super Yang-Mills theory in the planar limit by using the Eguchi-Kawai equivalence. The same matrix quantum mechanics is also used to provide a nonperturbative formulation of 4d N=1 super Yang-Mills theory on noncommutative space.

preprint2009arXiv

Large-N reduced models of supersymmetric quiver, Chern-Simons gauge theories and ABJM

Using the Eguchi-Kawai equivalence, we provide regularizations of supersymmetric quiver and Chern-Simons gauge theories which leave the supersymmetries unbroken. This allow us to study many interesting theories on a computer. As examples we construct large-$N$ reduced models of supersymmetric QCD with flavor and the ABJM model of multiple M2 branes.

preprint2007arXiv

Meta-Stable Supersymmetry Breaking in a Cooling Universe

We look at the recently proposed idea that susy breaking can be accomplished in a meta-stable vacuum. In the context of one of the simplest models (the Seiberg-dual of super-QCD), we address the following question: if we look at this theory as it cools from high temperature, is it at all possible that we can end up in a susy-breaking meta-stable vacuum? To get an idea about the answer, we look at the free energy of the system at high temperature. We conclude that the phase-structure of the free-energy as the temperature drops, is indeed such that there is a second order phase transition in the direction of the non-susy vacuum at a finite $T=T_c^Q$. On the other hand, the potential barrier in the direction of the susy vacuum is there all the way till $T \sim 0$.