-
Learning a Clinically-Relevant Concept Bottleneck for Lesion Detection in Breast Ultrasound
Authors:
Arianna Bunnell,
Yannik Glaser,
Dustin Valdez,
Thomas Wolfgruber,
Aleen Altamirano,
Carol Zamora González,
Brenda Y. Hernandez,
Peter Sadowski,
John A. Shepherd
Abstract:
Detecting and classifying lesions in breast ultrasound images is a promising application of artificial intelligence (AI) for reducing the burden of cancer in regions with limited access to mammography. Such AI systems are more likely to be useful in a clinical setting if their predictions can be explained to a radiologist. This work proposes an explainable AI model that provides interpretable pred…
▽ More
Detecting and classifying lesions in breast ultrasound images is a promising application of artificial intelligence (AI) for reducing the burden of cancer in regions with limited access to mammography. Such AI systems are more likely to be useful in a clinical setting if their predictions can be explained to a radiologist. This work proposes an explainable AI model that provides interpretable predictions using a standard lexicon from the American College of Radiology's Breast Imaging and Reporting Data System (BI-RADS). The model is a deep neural network featuring a concept bottleneck layer in which known BI-RADS features are predicted before making a final cancer classification. This enables radiologists to easily review the predictions of the AI system and potentially fix errors in real time by modifying the concept predictions. In experiments, a model is developed on 8,854 images from 994 women with expert annotations and histological cancer labels. The model outperforms state-of-the-art lesion detection frameworks with 48.9 average precision on the held-out testing set, and for cancer classification, concept intervention is shown to increase performance from 0.876 to 0.885 area under the receiver operating characteristic curve. Training and evaluation code is available at https://github.com/hawaii-ai/bus-cbm.
△ Less
Submitted 28 June, 2024;
originally announced July 2024.
-
Recurrence over Video Frames (RoVF) for the Re-identification of Meerkats
Authors:
Mitchell Rogers,
Kobe Knowles,
Gaël Gendron,
Shahrokh Heidari,
David Arturo Soriano Valdez,
Mihailo Azhar,
Padriac O'Leary,
Simon Eyre,
Michael Witbrock,
Patrice Delmas
Abstract:
Deep learning approaches for animal re-identification have had a major impact on conservation, significantly reducing the time required for many downstream tasks, such as well-being monitoring. We propose a method called Recurrence over Video Frames (RoVF), which uses a recurrent head based on the Perceiver architecture to iteratively construct an embedding from a video clip. RoVF is trained using…
▽ More
Deep learning approaches for animal re-identification have had a major impact on conservation, significantly reducing the time required for many downstream tasks, such as well-being monitoring. We propose a method called Recurrence over Video Frames (RoVF), which uses a recurrent head based on the Perceiver architecture to iteratively construct an embedding from a video clip. RoVF is trained using triplet loss based on the co-occurrence of individuals in the video frames, where the individual IDs are unavailable. We tested this method and various models based on the DINOv2 transformer architecture on a dataset of meerkats collected at the Wellington Zoo. Our method achieves a top-1 re-identification accuracy of $49\%$, which is higher than that of the best DINOv2 model ($42\%$). We found that the model can match observations of individuals where humans cannot, and our model (RoVF) performs better than the comparisons with minimal fine-tuning. In future work, we plan to improve these models by using pre-text tasks, apply them to animal behaviour classification, and perform a hyperparameter search to optimise the models further.
△ Less
Submitted 18 June, 2024;
originally announced June 2024.
-
Digital Evolution: Novo Nordisk's Shift to Ontology-Based Data Management
Authors:
Shawn Zheng Kai Tan,
Shounak Baksi,
Thomas Gade Bjerregaard,
Preethi Elangovan,
Thrishna Kuttikattu Gopalakrishnan,
Darko Hric,
Joffrey Joumaa,
Beidi Li,
Kashif Rabbani,
Santhosh Kannan Venkatesan,
Joshua Daniel Valdez,
Saritha Vettikunnel Kuriakose
Abstract:
Biomedical data is growing exponentially, and managing it is increasingly challenging. While Findable, Accessible, Interoperable and Reusable (FAIR) data principles provide guidance, their adoption has proven difficult, especially in larger enterprises like pharmaceutical companies. In this manuscript, we describe how we leverage an Ontology-Based Data Management (OBDM) strategy for digital transf…
▽ More
Biomedical data is growing exponentially, and managing it is increasingly challenging. While Findable, Accessible, Interoperable and Reusable (FAIR) data principles provide guidance, their adoption has proven difficult, especially in larger enterprises like pharmaceutical companies. In this manuscript, we describe how we leverage an Ontology-Based Data Management (OBDM) strategy for digital transformation in Novo Nordisk Research & Early Development. Here, we include both our technical blueprint and our approach for organizational change management. We further discuss how such an OBDM ecosystem plays a pivotal role in the organizations digital aspirations for data federation and discovery fuelled by artificial intelligence. Our aim for this paper is to share the lessons learned in order to foster dialogue with parties navigating similar waters while collectively advancing the efforts in the fields of data management, semantics and data driven drug discovery.
△ Less
Submitted 10 May, 2024; v1 submitted 8 May, 2024;
originally announced May 2024.
-
Behaviour Modelling of Social Animals via Causal Structure Discovery and Graph Neural Networks
Authors:
Gaël Gendron,
Yang Chen,
Mitchell Rogers,
Yiping Liu,
Mihailo Azhar,
Shahrokh Heidari,
David Arturo Soriano Valdez,
Kobe Knowles,
Padriac O'Leary,
Simon Eyre,
Michael Witbrock,
Gillian Dobbie,
Jiamou Liu,
Patrice Delmas
Abstract:
Better understanding the natural world is a crucial task with a wide range of applications. In environments with close proximity between humans and animals, such as zoos, it is essential to better understand the causes behind animal behaviour and what interventions are responsible for changes in their behaviours. This can help to predict unusual behaviours, mitigate detrimental effects and increas…
▽ More
Better understanding the natural world is a crucial task with a wide range of applications. In environments with close proximity between humans and animals, such as zoos, it is essential to better understand the causes behind animal behaviour and what interventions are responsible for changes in their behaviours. This can help to predict unusual behaviours, mitigate detrimental effects and increase the well-being of animals. There has been work on modelling the dynamics behind swarms of birds and insects but the complex social behaviours of mammalian groups remain less explored. In this work, we propose a method to build behavioural models using causal structure discovery and graph neural networks for time series. We apply this method to a mob of meerkats in a zoo environment and study its ability to predict future actions and model the behaviour distribution at an individual-level and at a group level. We show that our method can match and outperform standard deep learning architectures and generate more realistic data, while using fewer parameters and providing increased interpretability.
△ Less
Submitted 21 December, 2023;
originally announced December 2023.
-
Meerkat Behaviour Recognition Dataset
Authors:
Mitchell Rogers,
Gaël Gendron,
David Arturo Soriano Valdez,
Mihailo Azhar,
Yang Chen,
Shahrokh Heidari,
Caleb Perelini,
Padriac O'Leary,
Kobe Knowles,
Izak Tait,
Simon Eyre,
Michael Witbrock,
Patrice Delmas
Abstract:
Recording animal behaviour is an important step in evaluating the well-being of animals and further understanding the natural world. Current methods for documenting animal behaviour within a zoo setting, such as scan sampling, require excessive human effort, are unfit for around-the-clock monitoring, and may produce human-biased results. Several animal datasets already exist that focus predominant…
▽ More
Recording animal behaviour is an important step in evaluating the well-being of animals and further understanding the natural world. Current methods for documenting animal behaviour within a zoo setting, such as scan sampling, require excessive human effort, are unfit for around-the-clock monitoring, and may produce human-biased results. Several animal datasets already exist that focus predominantly on wildlife interactions, with some extending to action or behaviour recognition. However, there is limited data in a zoo setting or data focusing on the group behaviours of social animals. We introduce a large meerkat (Suricata Suricatta) behaviour recognition video dataset with diverse annotated behaviours, including group social interactions, tracking of individuals within the camera view, skewed class distribution, and varying illumination conditions. This dataset includes videos from two positions within the meerkat enclosure at the Wellington Zoo (Wellington, New Zealand), with 848,400 annotated frames across 20 videos and 15 unannotated videos.
△ Less
Submitted 20 June, 2023;
originally announced June 2023.
-
Insights into bootstrap percolation: Its equivalence with k-core percolation and the giant component
Authors:
M. A. Di Muro,
L. D. Valdez,
S. V. Buldyrev,
H. E. Stanley,
L. A. Braunstein
Abstract:
K-core and bootstrap percolation are widely studied models that have been used to represent and understand diverse deactivation and activation processes in natural and social systems. Since these models are considerably similar, it has been suggested in recent years that they could be complementary. In this manuscript we provide a rigorous analysis that shows that for any degree and threshold dist…
▽ More
K-core and bootstrap percolation are widely studied models that have been used to represent and understand diverse deactivation and activation processes in natural and social systems. Since these models are considerably similar, it has been suggested in recent years that they could be complementary. In this manuscript we provide a rigorous analysis that shows that for any degree and threshold distributions heterogeneous bootstrap percolation can be mapped into heterogeneous k-core percolation and vice versa, if the functionality thresholds in both processes satisfy a complementary relation. Another interesting problem in bootstrap and k-core percolation is the fraction of nodes belonging to their giant connected components $P_{\infty b}$ and $P_{\infty c}$, respectively. We solve this problem analytically for arbitrary randomly connected graphs and arbitrary threshold distributions, and we show that $P_{\infty b}$ and $P_{\infty c}$ are not complementary. Our theoretical results coincide with computer simulations in the limit of very large graphs. In bootstrap percolation, we show that when using the branching theory to compute the size of the giant component, we must consider two different types of links, which are related to distinct spanning branches of active nodes.
△ Less
Submitted 24 February, 2019; v1 submitted 9 November, 2018;
originally announced November 2018.
-
On Euclidean and Hermitian Self-Dual Cyclic Codes over $\mathbb{F}_{2^r}$
Authors:
Odessa D. Consorte,
Lilibeth D. Valdez
Abstract:
Cyclic and self-dual codes are important classes of codes in coding theory. Jia, Ling and Xing \cite{Jia} as well as Kai and Zhu \cite{Kai} proved that Euclidean self-dual cyclic codes of length $n$ over $\mathbb{F}_q$ exist if and only if $n$ is even and $q=2^r$, where $r$ is any positive integer. For $n$ and $q$ even, there always exists an $[n, \frac{n}{2}]$ self-dual cyclic code with generator…
▽ More
Cyclic and self-dual codes are important classes of codes in coding theory. Jia, Ling and Xing \cite{Jia} as well as Kai and Zhu \cite{Kai} proved that Euclidean self-dual cyclic codes of length $n$ over $\mathbb{F}_q$ exist if and only if $n$ is even and $q=2^r$, where $r$ is any positive integer. For $n$ and $q$ even, there always exists an $[n, \frac{n}{2}]$ self-dual cyclic code with generator polynomial $x^{\frac{n}{2}}+1$ called the \textit{trivial self-dual cyclic code}. In this paper we prove the existence of nontrivial self-dual cyclic codes of length $n=2^ν\cdot \bar{n}$, where $\bar{n}$ is odd, over $\mathbb{F}_{2^r}$ in terms of the existence of a nontrivial splitting $(Z, X_0, X_1)$ of $\mathbb{Z}_{\bar{n}}$ by $μ_{-1}$, where $Z, X_0,X_1$ are unions of $2^r$-cyclotomic cosets mod $\bar{n}.$ We also express the formula for the number of cyclic self-dual codes over $\mathbb{F}_{2^r}$ for each $n$ and $r$ in terms of the number of $2^r$-cyclotomic cosets in $X_0$ (or in $X_1$).
We also look at Hermitian self-dual cyclic codes and show properties which are analogous to those of Euclidean self-dual cyclic codes. That is, the existence of nontrivial Hermitian self-dual codes over $\mathbb{F}_{2^{2 \ell}}$ based on the existence of a nontrivial splitting $(Z, X_0, X_1)$ of $\mathbb{Z}_{\bar{n}}$ by $μ_{-2^\ell}$, where $Z, X_0,X_1$ are unions of $2^{2 \ell}$-cyclotomic cosets mod $\bar{n}.$ We also determine the lengths at which nontrivial Hermitian self-dual cyclic codes exist and the formula for the number of Hermitian self-dual cyclic codes for each $n$.
△ Less
Submitted 11 March, 2016;
originally announced March 2016.
-
Triple Point in Correlated Interdependent Networks
Authors:
L. D. Valdez,
P. A. Macri,
H. E. Stanley,
L. A. Braunstein
Abstract:
Many real-world networks depend on other networks, often in non-trivial ways, to maintain their functionality. These interdependent "networks of networks" are often extremely fragile. When a fraction $1-p$ of nodes in one network randomly fails, the damage propagates to nodes in networks that are interdependent and a dynamic failure cascade occurs that affects the entire system. We present dynamic…
▽ More
Many real-world networks depend on other networks, often in non-trivial ways, to maintain their functionality. These interdependent "networks of networks" are often extremely fragile. When a fraction $1-p$ of nodes in one network randomly fails, the damage propagates to nodes in networks that are interdependent and a dynamic failure cascade occurs that affects the entire system. We present dynamic equations for two interdependent networks that allow us to reproduce the failure cascade for an arbitrary pattern of interdependency. We study the "rich club" effect found in many real interdependent network systems in which the high-degree nodes are extremely interdependent, correlating a fraction $α$ of the higher degree nodes on each network. We find a rich phase diagram in the plane $p-α$, with a triple point reminiscent of the triple point of liquids that separates a non-functional phase from two functional phases.
△ Less
Submitted 19 November, 2013; v1 submitted 19 August, 2013;
originally announced August 2013.
-
Temporal percolation of a susceptible adaptive network
Authors:
L. D. Valdez,
P. A. Macri,
L. A. Braunstein
Abstract:
In the last decades, many authors have used the susceptible-infected-recovered model to study the impact of the disease spreading on the evolution of the infected individuals. However, few authors focused on the temporal unfolding of the susceptible individuals. In this paper, we study the dynamic of the susceptible-infected-recovered model in an adaptive network that mimics the transitory deactiv…
▽ More
In the last decades, many authors have used the susceptible-infected-recovered model to study the impact of the disease spreading on the evolution of the infected individuals. However, few authors focused on the temporal unfolding of the susceptible individuals. In this paper, we study the dynamic of the susceptible-infected-recovered model in an adaptive network that mimics the transitory deactivation of permanent social contacts, such as friendship and work-ship ties. Using an edge-based compartmental model and percolation theory, we obtain the evolution equations for the fraction susceptible individuals in the susceptible biggest component. In particular, we focus on how the individual's behavior impacts on the dilution of the susceptible network. We show that, as a consequence, the spreading of the disease slows down, protecting the biggest susceptible cluster by increasing the critical time at which the giant susceptible component is destroyed. Our theoretical results are fully supported by extensive simulations.
△ Less
Submitted 1 July, 2013; v1 submitted 28 September, 2012;
originally announced October 2012.
-
Temporal percolation of the susceptible network in an epidemic spreading
Authors:
L. D. Valdez,
P. A. Macri,
L. A. Braunstein
Abstract:
In this work, we study the evolution of the susceptible individuals during the spread of an epidemic modeled by the susceptible-infected-recovered (SIR) process spreading on the top of complex networks. Using an edge-based compartmental approach and percolation tools, we find that a time-dependent quantity $Φ_S(t)$, namely, the probability that a given neighbor of a node is susceptible at time…
▽ More
In this work, we study the evolution of the susceptible individuals during the spread of an epidemic modeled by the susceptible-infected-recovered (SIR) process spreading on the top of complex networks. Using an edge-based compartmental approach and percolation tools, we find that a time-dependent quantity $Φ_S(t)$, namely, the probability that a given neighbor of a node is susceptible at time $t$, is the control parameter of a node void percolation process involving those nodes on the network not-reached by the disease. We show that there exists a critical time $t_c$ above which the giant susceptible component is destroyed. As a consequence, in order to preserve a macroscopic connected fraction of the network composed by healthy individuals which guarantee its functionality, any mitigation strategy should be implemented before this critical time $t_c$. Our theoretical results are confirmed by extensive simulations of the SIR process.
△ Less
Submitted 23 September, 2012; v1 submitted 13 June, 2012;
originally announced June 2012.
-
Intermittent social distancing strategy for epidemic control
Authors:
L. D. Valdez,
P. A. Macri,
L. A. Braunstein
Abstract:
We study the critical effect of an intermittent social distancing strategy on the propagation of epidemics in adaptive complex networks. We characterize the effect of our strategy in the framework of the susceptible-infected-recovered model. In our model, based on local information, a susceptible individual interrupts the contact with an infected individual with a probability $σ$ and restores it a…
▽ More
We study the critical effect of an intermittent social distancing strategy on the propagation of epidemics in adaptive complex networks. We characterize the effect of our strategy in the framework of the susceptible-infected-recovered model. In our model, based on local information, a susceptible individual interrupts the contact with an infected individual with a probability $σ$ and restores it after a fixed time $t_{b}$. We find that, depending on the network topology, in our social distancing strategy there exists a cutoff threshold $σ_{c}$ beyond which the epidemic phase disappears. Our results are supported by a theoretical framework and extensive simulations of the model. Furthermore we show that this strategy is very efficient because it leads to a "susceptible herd behavior" that protects a large fraction of susceptibles individuals. We explain our results using percolation arguments.
△ Less
Submitted 20 March, 2012; v1 submitted 19 December, 2011;
originally announced December 2011.