-
A Multi-site Study of a Breast Density Deep Learning Model for Full-field Digital Mammography Images and Synthetic Mammography Images
Authors:
Thomas P. Matthews,
Sadanand Singh,
Brent Mombourquette,
Jason Su,
Meet P. Shah,
Stefano Pedemonte,
Aaron Long,
David Maffit,
Jenny Gurney,
Rodrigo Morales Hoil,
Nikita Ghare,
Douglas Smith,
Stephen M. Moore,
Susan C. Marks,
Richard L. Wahl
Abstract:
Purpose: To develop a Breast Imaging Reporting and Data System (BI-RADS) breast density deep learning (DL) model in a multi-site setting for synthetic two-dimensional mammography (SM) images derived from digital breast tomosynthesis exams using full-field digital mammography (FFDM) images and limited SM data.
Materials and Methods: A DL model was trained to predict BI-RADS breast density using F…
▽ More
Purpose: To develop a Breast Imaging Reporting and Data System (BI-RADS) breast density deep learning (DL) model in a multi-site setting for synthetic two-dimensional mammography (SM) images derived from digital breast tomosynthesis exams using full-field digital mammography (FFDM) images and limited SM data.
Materials and Methods: A DL model was trained to predict BI-RADS breast density using FFDM images acquired from 2008 to 2017 (Site 1: 57492 patients, 187627 exams, 750752 images) for this retrospective study. The FFDM model was evaluated using SM datasets from two institutions (Site 1: 3842 patients, 3866 exams, 14472 images, acquired from 2016 to 2017; Site 2: 7557 patients, 16283 exams, 63973 images, 2015 to 2019). Each of the three datasets were then split into training, validation, and test datasets. Adaptation methods were investigated to improve performance on the SM datasets and the effect of dataset size on each adaptation method is considered. Statistical significance was assessed using confidence intervals (CI), estimated by bootstrap**.
Results: Without adaptation, the model demonstrated substantial agreement with the original reporting radiologists for all three datasets (Site 1 FFDM: linearly-weighted $κ_w$ = 0.75 [95% CI: 0.74, 0.76]; Site 1 SM: $κ_w$ = 0.71 [95% CI: 0.64, 0.78]; Site 2 SM: $κ_w$ = 0.72 [95% CI: 0.70, 0.75]). With adaptation, performance improved for Site 2 (Site 1: $κ_w$ = 0.72 [95% CI: 0.66, 0.79], 0.71 vs 0.72, P = .80; Site 2: $κ_w$ = 0.79 [95% CI: 0.76, 0.81], 0.72 vs 0.79, P $<$ .001) using only 500 SM images from that site.
Conclusion: A BI-RADS breast density DL model demonstrated strong performance on FFDM and SM images from two institutions without training on SM images and improved using few SM images.
△ Less
Submitted 2 October, 2020; v1 submitted 23 January, 2020;
originally announced January 2020.
-
Period relations for Riemann surfaces with many automorphisms
Authors:
Luca Candelori,
Jack Fogliasso,
Christopher Marks,
Skip Moses
Abstract:
By employing the theory of vector-valued automorphic forms for non-unitarizable representations, we provide a new bound for the number of linear relations with algebraic coefficients between the periods of an algebraic Riemann surface with many automorphisms. The previous best-known general bound for this number was the genus of the Riemann surface, a result due to Wolfart. Our new bound significa…
▽ More
By employing the theory of vector-valued automorphic forms for non-unitarizable representations, we provide a new bound for the number of linear relations with algebraic coefficients between the periods of an algebraic Riemann surface with many automorphisms. The previous best-known general bound for this number was the genus of the Riemann surface, a result due to Wolfart. Our new bound significantly improves on this estimate, and it can be computed explicitly from the canonical representation of the Riemann surface. As observed by Shiga and Wolfart, this bound may then be used to estimate the dimension of the endomorphism algebra of the Jacobian of the Riemann surface. We demonstrate with a few examples how this improved bound allows one, in some instances, to actually compute the dimension of this endomorphism algebra, and to determine whether the Jacobian has complex multiplication.
△ Less
Submitted 16 December, 2018;
originally announced December 2018.
-
Ultrafast manipulation of mirror domain walls in a charge density wave
Authors:
Alfred Zong,
Xiaozhe Shen,
Anshul Kogar,
Linda Ye,
Carolyn Marks,
Debanjan Chowdhury,
Timm Rohwer,
Byron Freelon,
Stephen Weathersby,
Renkai Li,
Jie Yang,
Joseph Checkelsky,
Xijie Wang,
Nuh Gedik
Abstract:
Domain walls (DWs) are singularities in an ordered medium that often host exotic phenomena such as charge ordering, insulator-metal transition, or superconductivity. The ability to locally write and erase DWs is highly desirable, as it allows one to design material functionality by patterning DWs in specific configurations. We demonstrate such capability at room temperature in a charge density wav…
▽ More
Domain walls (DWs) are singularities in an ordered medium that often host exotic phenomena such as charge ordering, insulator-metal transition, or superconductivity. The ability to locally write and erase DWs is highly desirable, as it allows one to design material functionality by patterning DWs in specific configurations. We demonstrate such capability at room temperature in a charge density wave (CDW), a macroscopic condensate of electrons and phonons, in ultrathin 1T-TaS$_2$. A single femtosecond light pulse is shown to locally inject or remove mirror DWs in the CDW condensate, with probabilities tunable by pulse energy and temperature. Using time-resolved electron diffraction, we are able to simultaneously track anti-synchronized CDW amplitude oscillations from both the lattice and the condensate, where photo-injected DWs lead to a red-shifted frequency. Our demonstration of reversible DW manipulation may pave new ways for engineering correlated material systems with light.
△ Less
Submitted 25 October, 2018;
originally announced October 2018.
-
Building a Location-Based Set of Social Media Users
Authors:
Christopher Marks,
Tauhid Zaman
Abstract:
In many instances one may want to gain situational awareness in an environment by monitoring the content of local social media users. Often the challenge is how to build a set of users from a target location. Here we introduce a method for building such a set of users by using an \emph{expand-classify} approach which begins with a small set of seed users from the target location and then iterative…
▽ More
In many instances one may want to gain situational awareness in an environment by monitoring the content of local social media users. Often the challenge is how to build a set of users from a target location. Here we introduce a method for building such a set of users by using an \emph{expand-classify} approach which begins with a small set of seed users from the target location and then iteratively collects their neighbors and then classifies their locations. We perform this classification using maximum likelihood estimation on a factor graph model which incorporates features of the user profile and also social network connections. We show that maximum likelihood estimation reduces to solving a minimum cut problem on an appropriately defined graph. We are able to obtain several thousand users within a few hours for many diverse locations using our approach. Using geo-located data, we find that our approach typically achieves good accuracy for population centers with less than 500,000 inhabitants, while for larger cities performance degrades somewhat. We also find that our approach is able to collect many more users with higher accuracy than existing search methods. Finally, we show that by studying the content of location specific users obtained with our approach, we can identify the onset of significant social unrest in locations such as the Philippines.
△ Less
Submitted 4 November, 2017;
originally announced November 2017.
-
Indecomposable vector-valued modular forms and periods of modular curves
Authors:
Luca Candelori,
Tucker Hartland,
Christopher Marks,
Diego Yepez
Abstract:
We classify the three-dimensional representations of the modular group that are reducible but indecomposable, and their associated spaces of holomorphic vector-valued modular forms. We then demonstrate how such representations may be employed to compute periods of modular curves. This technique obviates the use of Hecke operators, and therefore provides a method for studying noncongruence modular…
▽ More
We classify the three-dimensional representations of the modular group that are reducible but indecomposable, and their associated spaces of holomorphic vector-valued modular forms. We then demonstrate how such representations may be employed to compute periods of modular curves. This technique obviates the use of Hecke operators, and therefore provides a method for studying noncongruence modular curves as well as congruence.
△ Less
Submitted 14 October, 2017; v1 submitted 6 July, 2017;
originally announced July 2017.
-
Intertwining operators and vector-valued modular forms for minimal models
Authors:
Matthew Krauel,
Christopher Marks
Abstract:
Using the language of vertex operator algebras (VOAs) and vector-valued modular forms we study the modular group representations and spaces of 1-point functions associated to intertwining operators for Virasoro minimal model VOAs. We examine all representations of dimension less than four associated to irreducible modules for minimal models, and determine when the kernel of these representations i…
▽ More
Using the language of vertex operator algebras (VOAs) and vector-valued modular forms we study the modular group representations and spaces of 1-point functions associated to intertwining operators for Virasoro minimal model VOAs. We examine all representations of dimension less than four associated to irreducible modules for minimal models, and determine when the kernel of these representations is a congruence or noncongruence subgroup of the modular group. Arithmetic criteria are given on the indexing of the irreducible modules for minimal models that imply the associated modular group representation has a noncongruence kernel, independent of the dimension of the representation. The algebraic structure of the spaces of 1-point functions for intertwining operators is also studied, via a comparison with the associated spaces of holomorphic vector-valued modular forms.
△ Less
Submitted 7 December, 2016;
originally announced December 2016.
-
Finding Online Extremists in Social Networks
Authors:
Jytte Klausen,
Christopher Marks,
Tauhid Zaman
Abstract:
Online extremists in social networks pose a new form of threat to the general public. These extremists range from cyberbullies who harass innocent users to terrorist organizations such as the Islamic State of Iraq and Syria (ISIS) that use social networks to recruit and incite violence. Currently social networks suspend the accounts of such extremists in response to user complaints. The challenge…
▽ More
Online extremists in social networks pose a new form of threat to the general public. These extremists range from cyberbullies who harass innocent users to terrorist organizations such as the Islamic State of Iraq and Syria (ISIS) that use social networks to recruit and incite violence. Currently social networks suspend the accounts of such extremists in response to user complaints. The challenge is that these extremist users simply create new accounts and continue their activities. In this work we present a new set of operational capabilities to deal with the threat posed by online extremists in social networks.
Using data from several hundred thousand extremist accounts on Twitter, we develop a behavioral model for these users, in particular what their accounts look like and who they connect with. This model is used to identify new extremist accounts by predicting if they will be suspended for extremist activity. We also use this model to track existing extremist users as they create new accounts by identifying if two accounts belong to the same user. Finally, we present a model for searching the social network to efficiently find suspended users' new accounts based on a variant of the classic Polya's urn setup. We find a simple characterization of the optimal search policy for this model under fairly general conditions. Our urn model and main theoretical results generalize easily to search problems in other fields.
△ Less
Submitted 19 October, 2016;
originally announced October 2016.
-
A Multi-Urn Model for Network Search
Authors:
Christopher E. Marks,
Tauhid Zaman
Abstract:
We consider the problem of finding a specific target individual hiding in a social network. We propose a method for network vertex search that looks for the target vertex by sequentially examining the neighbors of a set of "known" vertices to which the target vertex may have connected. The objective is to find the target vertex as quickly as possible from amongst the neighbors of the known vertice…
▽ More
We consider the problem of finding a specific target individual hiding in a social network. We propose a method for network vertex search that looks for the target vertex by sequentially examining the neighbors of a set of "known" vertices to which the target vertex may have connected. The objective is to find the target vertex as quickly as possible from amongst the neighbors of the known vertices. We model this type of search as successively drawing marbles from a set of urns, where each urn represents one of the known vertices and the marbles in each urn represent the respective vertex's neighbors. Using a dynamic programming approach, we analyze this model and show that there is always an optimal "block" policy, in which all of the neighbors of a known vertex are examined before moving on to another vertex. Surprisingly, this block policy result holds for arbitrary dependencies in the connection probabilities of the target vertex and known vertices. Furthermore, we give precise characterizations of the optimal block policy in two specific cases: (1) when the connections between the known vertices and the target vertex are independent, and (2) when the target vertex is connected to at most one known vertex. Finally, we provide some general monotonicity properties and discuss the relevance of our findings in social media and other applications.
△ Less
Submitted 24 August, 2016;
originally announced August 2016.
-
Fourier coefficients of three-dimensional vector-valued modular forms
Authors:
Christopher Marks
Abstract:
A thorough analysis is made of the Fourier coefficients for vector-valued modular forms associated to three-dimensional irreducible representations of the modular group. In particular, the following statement is verified for all but a finite number of equivalence classes: if a vector-valued modular form associated to such a representation has rational Fourier coefficients, then these coefficients…
▽ More
A thorough analysis is made of the Fourier coefficients for vector-valued modular forms associated to three-dimensional irreducible representations of the modular group. In particular, the following statement is verified for all but a finite number of equivalence classes: if a vector-valued modular form associated to such a representation has rational Fourier coefficients, then these coefficients have "unbounded denominators", i.e. there is a prime number p, depending on the representation, which occurs to an arbitrarily high power in the denominators of the coefficients. This provides a verification in the three-dimensional setting of a generalization of a long-standing conjecture about noncongruence modular forms.
△ Less
Submitted 31 March, 2015; v1 submitted 24 January, 2012;
originally announced January 2012.
-
Irreducible vector-valued modular forms of dimension less than six
Authors:
Christopher Marks
Abstract:
An algebraic classification is given for spaces of holomorphic vector-valued modular forms of arbitrary real weight and multiplier system, associated to irreducible, T-unitarizable representations of the full modular group, of dimension less than six. For representations of dimension less than four, it is shown that the associated space of vector-valued modular forms is a cyclic module over a cert…
▽ More
An algebraic classification is given for spaces of holomorphic vector-valued modular forms of arbitrary real weight and multiplier system, associated to irreducible, T-unitarizable representations of the full modular group, of dimension less than six. For representations of dimension less than four, it is shown that the associated space of vector-valued modular forms is a cyclic module over a certain skew polynomial ring of differential operators. For dimensions four and five, a complete list of possible Hilbert-Poincare series is given, using the fact that the space of vector-valued modular forms is a free module over the ring of classical modular forms for the full modular group. A mild restriction is then placed on the class of representation considered in these dimensions, and this again yields an explicit determination of the associated Hilbert-Poincare series.
△ Less
Submitted 26 January, 2012; v1 submitted 18 April, 2010;
originally announced April 2010.
-
Classification of Vector-Valued modular Forms of Dimension less than Six
Authors:
Christopher Marks
Abstract:
The first half of this dissertation reviews the basic notion of vector-valued modular forms and its connection to differential equations. The main purpose of the dissertation is to classify spaces of vector-valued modular forms associated to irreducible, T-unitarizable representations of the full modular group, of dimension less than six. Given such a representation, it is shown that the associate…
▽ More
The first half of this dissertation reviews the basic notion of vector-valued modular forms and its connection to differential equations. The main purpose of the dissertation is to classify spaces of vector-valued modular forms associated to irreducible, T-unitarizable representations of the full modular group, of dimension less than six. Given such a representation, it is shown that the associated graded complex linear space of vector-valued modular forms is a free module over the ring of integral weight modular forms for the full modular group, whose rank is equal to the dimension of the given representation. An explicit basis is computed for the module structure in each case, and this basis is used to compute the Hilbert-Poincare series associated to each graded space.
△ Less
Submitted 22 March, 2010;
originally announced March 2010.
-
Structure of the module of vector-valued modular forms
Authors:
Christopher Marks,
Geoffrey Mason
Abstract:
Let $V$ be a representation of the modular group $Γ$ of dimension $p$. We show that the $\mathbb{Z}$-graded space $\mathcal{H}(V)$ of holomorphic vector-valued modular forms associated to $V$ is a free module of rank $p$ over the algebra $\mathcal{M}$ of classical holomorphic modular forms. We study the nature of $\mathcal{H}$ considered as a functor from $Γ$-modules to graded $\mathcal{M}$-latt…
▽ More
Let $V$ be a representation of the modular group $Γ$ of dimension $p$. We show that the $\mathbb{Z}$-graded space $\mathcal{H}(V)$ of holomorphic vector-valued modular forms associated to $V$ is a free module of rank $p$ over the algebra $\mathcal{M}$ of classical holomorphic modular forms. We study the nature of $\mathcal{H}$ considered as a functor from $Γ$-modules to graded $\mathcal{M}$-lattices and give some applications, including the calculation of the Hilbert-Poincaré of $\mathcal{H}(V)$ in some cases.
△ Less
Submitted 27 January, 2009;
originally announced January 2009.
