Showing 1–38 of 38 results for author: Edwards, M

HCCL: Collective Communication for Meta Training and Inference Accelerators

Authors: Wesley Bland, Tiago Antunes, Lars Paul Huse, Chidambaram Muthu, Adel Abouchaev, Rabib Alam, Abdullah Alperen, Alexey Andronov, Jose Anto Akkara, Vineet Badhwar, Pavan Balaji, Daniel Berkovitch, Bartosz Bogdanski, Shmeelok Chakraborty, Sungjun Cho, John Choi, James Custer, Rodrigo De Castro, Nguyen Dinh Pham, Matthew Edwards, Kristian Evensen, Evan Ezell, Alex Finestead, Seth Goldstein, Prankur Gupta , et al. (41 additional authors not shown)

Abstract: We present HCCL, a collective communication library co-designed with Meta's MTIA 300 accelerator, the first Meta chip to integrate backend networking directly on chip package. MTIA 300 includes dedicated message engines (MEs) with near-memory compute (NMC) that fully offload collective execution from the compute grid, enabling large overlap between computation and communication. HCCL uses a compil… ▽ More We present HCCL, a collective communication library co-designed with Meta's MTIA 300 accelerator, the first Meta chip to integrate backend networking directly on chip package. MTIA 300 includes dedicated message engines (MEs) with near-memory compute (NMC) that fully offload collective execution from the compute grid, enabling large overlap between computation and communication. HCCL uses a compiled communication model in which the host generates a complete description of each collective including dependencies. We describe the control and data path architecture, topology-aware algorithm selection across MTIA 300's asymmetric scale-up and scale-out network, and optimizations for both training and inference workloads. For training, HCCL achieves up to 940 GB/s on intra-rack collectives while introducing less than 0.5% degradation to concurrent compute throughput. For inference, we leverage one-sided communication primitives that bypass the scheduling path to minimize collective latency and describe collective designs that improve compute-communication pipelining for latency-sensitive workloads. △ Less

Submitted 31 July, 2026; originally announced August 2026.

Comments: 12 pages, 17 figures, to be published in the proceedings of "SC '26: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis"

Nonslop: A Gamified Experiment in Human-AI Collaborative Writing

Authors: Maria Edwards, Julian Togelius

Abstract: The rapid proliferation of large language models (LLMs) raises critical questions about human creativity and individual expression in an era of AI-assisted creation. When do humans adopt AI suggestions, and what are the implications for individual voice? This study examines these questions through a gamified writing exercise where 74 participants (214 responses) replied to prompts while AI-gener… ▽ More The rapid proliferation of large language models (LLMs) raises critical questions about human creativity and individual expression in an era of AI-assisted creation. When do humans adopt AI suggestions, and what are the implications for individual voice? This study examines these questions through a gamified writing exercise where 74 participants (214 responses) replied to prompts while AI-generated word suggestions were available as they wrote. The game simulates a dystopian future in which an AI is attempting to learn from what remains of human individuality, and disincentivizes AI-like writing. In doing so, it attempts to create conditions that reveal authentic user preferences rather than default behaviors, such as accepting a readily available AI-generated suggestion. Note that this is a deliberate inversion of the "helpful assistant" design pattern; the system is explicitly forbidding you from accepting AI suggestions. We analyze user behavior patterns across different task types, user behaviors, and response characteristics to understand the factors influencing human-AI interaction in creative tasks. The study focuses on when users choose to maintain creative autonomy versus violating the rules of the game and accepting AI assistance. It also explores how these choices relate to response patterns, task characteristics, and user behavior. This gamified approach offers both a framework for studying authentic human-AI interaction and a provocative lens for understanding the tension between efficiency and authenticity in AI-augmented creativity. △ Less

Submitted 10 June, 2026; originally announced June 2026.

Comments: Accepted at the 2026 IEEE Conference on Games (CoG 2026); to be published in the conference proceedings. Camera-ready version

Agentic AI in Engineering and Manufacturing: Industry Perspectives on Utility, Adoption, Challenges, and Opportunities

Authors: Kristen M. Edwards, Maxwell Bauer, Claire Jacquillat, A. John Hart, Faez Ahmed

Abstract: This work examines how AI, especially agentic systems, is being adopted in engineering and manufacturing workflows, what value it provides today, and what is needed for broader deployment. This is an exploratory and qualitative state-of-practice study grounded in over 30 interviews across four stakeholder groups (large enterprises, small/medium firms, AI developers, and CAD/CAM/CAE vendors). We fi… ▽ More This work examines how AI, especially agentic systems, is being adopted in engineering and manufacturing workflows, what value it provides today, and what is needed for broader deployment. This is an exploratory and qualitative state-of-practice study grounded in over 30 interviews across four stakeholder groups (large enterprises, small/medium firms, AI developers, and CAD/CAM/CAE vendors). We find that near-term AI gains cluster around structured, repetitive work and data-intensive synthesis, while higher-value agentic gains come from orchestrating multi-step workflows across tools. Adoption is constrained less by model capability than by fragmented and machine-unfriendly data, stringent security and regulatory requirements, and limited API-accessible legacy toolchains. Reliability, verification, and auditability are central requirements for adoption, driving human-in-the-loop frameworks and governance aligned with existing engineering reviews. Beyond technical barriers there are also organizational ones: a persistent AI literacy gap, cultural heterogeneity, and governance structures that have not yet caught up with agentic capabilities. Together, the findings point to a staged progression of AI utility from low-consequence assistance toward higher-order automation, as trust, infrastructure, and verification mature. This highlights key breakthroughs needed, including integration with traditional engineering tools and data types, robust verification frameworks, and improved spatial and physical reasoning. △ Less

Submitted 19 March, 2026; originally announced April 2026.

Comments: Funding and support from the MIT Initiative for New Manufacturing

Integrating Genomics into Multimodal EHR Foundation Models

Authors: Jonathan Amar, Edward Liu, Alessandra Breschi, Liangliang Zhang, Pouya Kheradpour, Sylvia Li, Lisa Soleymani Lehmann, Alessandro Giulianelli, Matt Edwards, Yugang Jia, David Nola, Raghav Mani, Pankaj Vats, Jesse Tetreault, T. J. Chen, Cory Y. McLean

Abstract: This paper introduces an innovative Electronic Health Record (EHR) foundation model that integrates Polygenic Risk Scores (PRS) as a foundational data modality, moving beyond traditional EHR-only approaches to build more holistic health profiles. Leveraging the extensive and diverse data from the All of Us (AoU) Research Program, this multimodal framework aims to learn complex relationships betwee… ▽ More This paper introduces an innovative Electronic Health Record (EHR) foundation model that integrates Polygenic Risk Scores (PRS) as a foundational data modality, moving beyond traditional EHR-only approaches to build more holistic health profiles. Leveraging the extensive and diverse data from the All of Us (AoU) Research Program, this multimodal framework aims to learn complex relationships between clinical data and genetic predispositions. The methodology extends advancements in generative AI to the EHR foundation model space, enhancing predictive capabilities and interpretability. Evaluation on AoU data demonstrates the model's predictive value for the onset of various conditions, particularly Type 2 Diabetes (T2D), and illustrates the interplay between PRS and EHR data. The work also explores transfer learning for custom classification tasks, showcasing the architecture's versatility and efficiency. This approach is pivotal for unlocking new insights into disease prediction, proactive health management, risk stratification, and personalized treatment strategies, laying the groundwork for more personalized, equitable, and actionable real-world evidence generation in healthcare. △ Less

Submitted 14 November, 2025; v1 submitted 24 October, 2025; originally announced October 2025.

Compiling a Q# Subset to QASM 3.0 in TypeScript via a JSON Based IR

Authors: Marcus Edwards

Abstract: We implement a compile toolchain from Q# to QASM 3.0 including a full-featured lexer and parser implementation, as well as a compiler that supports a subset of Q# features. The lexer, parser and compiler are shown to work with various input Q# programs and the implementation is compared against existing Q# compile tools. Unlike the Microsoft implementation of the official Q# compile toolchain, our… ▽ More We implement a compile toolchain from Q# to QASM 3.0 including a full-featured lexer and parser implementation, as well as a compiler that supports a subset of Q# features. The lexer, parser and compiler are shown to work with various input Q# programs and the implementation is compared against existing Q# compile tools. Unlike the Microsoft implementation of the official Q# compile toolchain, our implementation is written in TypeScript in order to port functionality to web environments. △ Less

Submitted 16 July, 2025; v1 submitted 29 June, 2025; originally announced June 2025.

Productive Quantum Programming Needs Better Abstract Machines

Authors: Santiago Núñez-Corrales, Olivia Di Matteo, John Dumbell, Marcus Edwards, Edoardo Giusto, Scott Pakin, Vlad Stirbu

Abstract: An effective, accessible abstraction hierarchy has made using and programming computers possible for people across all disciplines. Establishing such a hierarchy for quantum programming is an outstanding challenge, especially due to a proliferation of different conventions and the rapid pace of innovation. One critical portion of the hierarchy is the abstract machine, the layer that separates a pr… ▽ More An effective, accessible abstraction hierarchy has made using and programming computers possible for people across all disciplines. Establishing such a hierarchy for quantum programming is an outstanding challenge, especially due to a proliferation of different conventions and the rapid pace of innovation. One critical portion of the hierarchy is the abstract machine, the layer that separates a programmer's mental model of the hardware from its physical realization. Drawing on historical parallels in classical computing, we explain why having the "right" quantum abstract machine (QAM) is essential for making progress in the field and propose a novel framework for evaluating QAMs based on a set of desirable criteria. These criteria capture aspects of a QAM such as universality, compactness, expressiveness, and composability, which aid in the representation of quantum programs. By defining this framework we take steps toward defining an optimal QAM. We further apply our framework to survey the landscape of existing proposals, draw comparisons, and assess them based on our criteria. While these proposals share many common strengths, we find that each falls short of our ideal. Our framework and our findings set a direction for subsequent efforts to define a future QAM that is both straightforward to map to a variety of quantum computers, and provides a stable abstraction for quantum software development. △ Less

Submitted 17 April, 2025; originally announced May 2025.

Comments: 11 pages, 2 figures, 1 table

AI Judges in Design: Statistical Perspectives on Achieving Human Expert Equivalence With Vision-Language Models

Authors: Kristen M. Edwards, Farnaz Tehranchi, Scarlett R. Miller, Faez Ahmed

Abstract: The subjective evaluation of early stage engineering designs, such as conceptual sketches, traditionally relies on human experts. However, expert evaluations are time-consuming, expensive, and sometimes inconsistent. Recent advances in vision-language models (VLMs) offer the potential to automate design assessments, but it is crucial to ensure that these AI ``judges'' perform on par with human exp… ▽ More The subjective evaluation of early stage engineering designs, such as conceptual sketches, traditionally relies on human experts. However, expert evaluations are time-consuming, expensive, and sometimes inconsistent. Recent advances in vision-language models (VLMs) offer the potential to automate design assessments, but it is crucial to ensure that these AI ``judges'' perform on par with human experts. However, no existing framework assesses expert equivalence. This paper introduces a rigorous statistical framework to determine whether an AI judge's ratings match those of human experts. We apply this framework in a case study evaluating four VLM-based judges on key design metrics (uniqueness, creativity, usefulness, and drawing quality). These AI judges employ various in-context learning (ICL) techniques, including uni- vs. multimodal prompts and inference-time reasoning. The same statistical framework is used to assess three trained novices for expert-equivalence. Results show that the top-performing AI judge, using text- and image-based ICL with reasoning, achieves expert-level agreement for uniqueness and drawing quality and outperforms or matches trained novices across all metrics. In 6/6 runs for both uniqueness and creativity, and 5/6 runs for both drawing quality and usefulness, its agreement with experts meets or exceeds that of the majority of trained novices. These findings suggest that reasoning-supported VLM models can achieve human-expert equivalence in design evaluation. This has implications for scaling design evaluation in education and practice, and provides a general statistical framework for validating AI judges in other domains requiring subjective content evaluation. △ Less

Submitted 1 April, 2025; originally announced April 2025.

Comments: 21 pages, 8 tables, 6 figures, 8 tables in the appendix

Enabling the Verification and Formalization of Hybrid Quantum-Classical Computing with OpenQASM 3.0 compatible QASM-TS 2.0

Authors: Sean Kim, Marcus Edwards

Abstract: The unique features of the hybrid quantum-classical computing model implied by the specification of OpenQASM 3.0 motivate new approaches to quantum program verification. We implement and thoroughly test a QASM 3.0 parser in TypeScript to enable implementations of verification and validation software, compilers, and more. We aim to help the community to formalize the logic of hybrid quantum-classic… ▽ More The unique features of the hybrid quantum-classical computing model implied by the specification of OpenQASM 3.0 motivate new approaches to quantum program verification. We implement and thoroughly test a QASM 3.0 parser in TypeScript to enable implementations of verification and validation software, compilers, and more. We aim to help the community to formalize the logic of hybrid quantum-classical computing by providing tools that may help with such efforts. △ Less

Submitted 22 September, 2025; v1 submitted 17 December, 2024; originally announced December 2024.

Comments: Accepted version

Journal ref: Journal of Open Source Software (2025), 10(113), 8696

Time-to-Lie: Identifying Industrial Control System Honeypots Using the Internet Control Message Protocol

Authors: Jacob Williams, Matthew Edwards, Joseph Gardiner

Abstract: The convergence of information and operational technology networks has created previously unforeseen security issues. To address these issues, both researchers and practitioners have integrated threat intelligence methods into the security operations of converged networks, with some of the most valuable tools being honeypots that imitate industrial control systems (ICS). However, the development a… ▽ More The convergence of information and operational technology networks has created previously unforeseen security issues. To address these issues, both researchers and practitioners have integrated threat intelligence methods into the security operations of converged networks, with some of the most valuable tools being honeypots that imitate industrial control systems (ICS). However, the development and deployment of such honeypots is a process rich with pitfalls, which can lead to undiagnosed weaknesses in the threat intelligence being gathered. This paper presents a side-channel method of covertly identifying ICS honeypots using the time-to-live (TTL) values of target devices. We show that many ICS honeypots can be readily identified, via minimal interactions, using only basic networking tools. In a study of over 8,000 devices presenting as ICS systems, we detail how our method compares to an existing honeypot detection approach, and outline what our methodology reveals about the current population of live ICS honeypots. In demonstrating our method, this study aims to raise awareness of the viability of the TTL heuristic and the prevalence of its misconfiguration despite its presence in literature. △ Less

Submitted 23 October, 2024; originally announced October 2024.

Comments: 11 pages, 2 listings, 5 tables, 6 figures

Sketch2Prototype: Rapid Conceptual Design Exploration and Prototyping with Generative AI

Authors: Kristen M. Edwards, Brandon Man, Faez Ahmed

Abstract: Sketch2Prototype is an AI-based framework that transforms a hand-drawn sketch into a diverse set of 2D images and 3D prototypes through sketch-to-text, text-to-image, and image-to-3D stages. This framework, shown across various sketches, rapidly generates text, image, and 3D modalities for enhanced early-stage design exploration. We show that using text as an intermediate modality outperforms dire… ▽ More Sketch2Prototype is an AI-based framework that transforms a hand-drawn sketch into a diverse set of 2D images and 3D prototypes through sketch-to-text, text-to-image, and image-to-3D stages. This framework, shown across various sketches, rapidly generates text, image, and 3D modalities for enhanced early-stage design exploration. We show that using text as an intermediate modality outperforms direct sketch-to-3D baselines for generating diverse and manufacturable 3D models. We find limitations in current image-to-3D techniques, while noting the value of the text modality for user-feedback and iterative design augmentation. △ Less

Submitted 25 March, 2024; originally announced May 2024.

Comments: 10 pages, 7 figures

Helpful or Harmful? Exploring the Efficacy of Large Language Models for Online Grooming Prevention

Authors: Ellie Prosser, Matthew Edwards

Abstract: Powerful generative Large Language Models (LLMs) are becoming popular tools amongst the general public as question-answering systems, and are being utilised by vulnerable groups such as children. With children increasingly interacting with these tools, it is imperative for researchers to scrutinise the safety of LLMs, especially for applications that could lead to serious outcomes, such as online… ▽ More Powerful generative Large Language Models (LLMs) are becoming popular tools amongst the general public as question-answering systems, and are being utilised by vulnerable groups such as children. With children increasingly interacting with these tools, it is imperative for researchers to scrutinise the safety of LLMs, especially for applications that could lead to serious outcomes, such as online child safety queries. In this paper, the efficacy of LLMs for online grooming prevention is explored both for identifying and avoiding grooming through advice generation, and the impact of prompt design on model performance is investigated by varying the provided context and prompt specificity. In results reflecting over 6,000 LLM interactions, we find that no models were clearly appropriate for online grooming prevention, with an observed lack of consistency in behaviours, and potential for harmful answer generation, especially from open-source models. We outline where and how models fall short, providing suggestions for improvement, and identify prompt designs that heavily altered model performance in troubling ways, with findings that can be used to inform best practice usage guides. △ Less

Submitted 14 March, 2024; originally announced March 2024.

Towards Understanding Underwater Weather Events in Rivers Using Autonomous Surface Vehicles

Authors: Alice K. Li, Yue Mao, Sandeep Manjanna, Sixuan Liu, Jasleen Dhanoa, Bharg Mehta, Victoria M. Edwards, Fernando Cladera Ojeda, Maël Le Men, Eric Sigg, Hugo N. Ulloa, Douglas J. Jerolmack, M. Ani Hsieh

Abstract: Climate change has increased the frequency and severity of extreme weather events such as hurricanes and winter storms. The complex interplay of floods with tides, runoff, and sediment creates additional hazards -- including erosion and the undermining of urban infrastructure -- consequently impacting the health of our rivers and ecosystems. Observations of these underwater phenomena are rare, bec… ▽ More Climate change has increased the frequency and severity of extreme weather events such as hurricanes and winter storms. The complex interplay of floods with tides, runoff, and sediment creates additional hazards -- including erosion and the undermining of urban infrastructure -- consequently impacting the health of our rivers and ecosystems. Observations of these underwater phenomena are rare, because satellites and sensors mounted on aerial vehicles cannot penetrate the murky waters. Autonomous Surface Vehicles (ASVs) provides a means to track and map these complex and dynamic underwater phenomena. This work highlights preliminary results of high-resolution data gathering with ASVs, equipped with a suite of sensors capable of measuring physical and chemical parameters of the river. Measurements were acquired along the lower Schuylkill River in the Philadelphia area at high-tide and low-tide conditions. The data will be leveraged to improve our understanding of changes in bathymetry due to floods; the dynamics of mixing and stagnation zones and their impact on water quality; and the dynamics of suspension and resuspension of fine sediment. The data will also provide insight into the development of adaptive sampling strategies for ASVs that can maximize the information gain for future field experiments. △ Less

Submitted 21 December, 2023; originally announced December 2023.

Comments: Published to IEEE OCEANS 2022

From Concept to Manufacturing: Evaluating Vision-Language Models for Engineering Design

Authors: Cyril Picard, Kristen M. Edwards, Anna C. Doris, Brandon Man, Giorgio Giannone, Md Ferdous Alam, Faez Ahmed

Abstract: Engineering design is undergoing a transformative shift with the advent of AI, marking a new era in how we approach product, system, and service planning. Large language models have demonstrated impressive capabilities in enabling this shift. Yet, with text as their only input modality, they cannot leverage the large body of visual artifacts that engineers have used for centuries and are accustome… ▽ More Engineering design is undergoing a transformative shift with the advent of AI, marking a new era in how we approach product, system, and service planning. Large language models have demonstrated impressive capabilities in enabling this shift. Yet, with text as their only input modality, they cannot leverage the large body of visual artifacts that engineers have used for centuries and are accustomed to. This gap is addressed with the release of multimodal vision-language models (VLMs), such as GPT-4V, enabling AI to impact many more types of tasks. Our work presents a comprehensive evaluation of VLMs across a spectrum of engineering design tasks, categorized into four main areas: Conceptual Design, System-Level and Detailed Design, Manufacturing and Inspection, and Engineering Education Tasks. Specifically in this paper, we assess the capabilities of two VLMs, GPT-4V and LLaVA 1.6 34B, in design tasks such as sketch similarity analysis, CAD generation, topology optimization, manufacturability assessment, and engineering textbook problems. Through this structured evaluation, we not only explore VLMs' proficiency in handling complex design challenges but also identify their limitations in complex engineering design applications. Our research establishes a foundation for future assessments of vision language models. It also contributes a set of benchmark testing datasets, with more than 1000 queries, for ongoing advancements and applications in this field. △ Less

Submitted 9 December, 2024; v1 submitted 21 November, 2023; originally announced November 2023.

Text generation for dataset augmentation in security classification tasks

Authors: Alexander P. Welsh, Matthew Edwards

Abstract: Security classifiers, designed to detect malicious content in computer systems and communications, can underperform when provided with insufficient training data. In the security domain, it is often easy to find samples of the negative (benign) class, and challenging to find enough samples of the positive (malicious) class to train an effective classifier. This study evaluates the application of n… ▽ More Security classifiers, designed to detect malicious content in computer systems and communications, can underperform when provided with insufficient training data. In the security domain, it is often easy to find samples of the negative (benign) class, and challenging to find enough samples of the positive (malicious) class to train an effective classifier. This study evaluates the application of natural language text generators to fill this data gap in multiple security-related text classification tasks. We describe a variety of previously-unexamined language-model fine-tuning approaches for this purpose and consider in particular the impact of disproportionate class-imbalances in the training set. Across our evaluation using three state-of-the-art classifiers designed for offensive language detection, review fraud detection, and SMS spam detection, we find that models trained with GPT-3 data augmentation strategies outperform both models trained without augmentation and models trained using basic data augmentation strategies already in common usage. In particular, we find substantial benefits for GPT-3 data augmentation strategies in situations with severe limitations on known positive-class samples. △ Less

Submitted 22 October, 2023; originally announced October 2023.

Three Quantum Programming Language Parser Implementations for the Web

Authors: Marcus Edwards

Abstract: IBM has developed a quantum assembly (QASM) language particular to gate model quantum computing since 2017 [CBSG17]. Version 3.0 which adds timing, pulse control, and gate modifiers is currently undergoing finalization in 2023 [CJA+21]. In a similar vein, Pakin of Los Alamos National Laboratory published a quantum macro assembler (QMASM) for D-Wave quantum annealers in 2016 [Pak16]. This assembler… ▽ More IBM has developed a quantum assembly (QASM) language particular to gate model quantum computing since 2017 [CBSG17]. Version 3.0 which adds timing, pulse control, and gate modifiers is currently undergoing finalization in 2023 [CJA+21]. In a similar vein, Pakin of Los Alamos National Laboratory published a quantum macro assembler (QMASM) for D-Wave quantum annealers in 2016 [Pak16]. This assembler specifically targets quantum annealers like D-Wave's. A comparable technology that targets continuous-variable (CV) quantum computing is the Blackbird language developed by Xanadu since 2018 [KIQ+19]. We implement parsers for each of these languages in TypeScript with a singular approach. In the cases of Blackbird and QMASM these are the first parser implementations that are web compatible and so bring these languages to a new audience and to new runtimes. This makes the parsing and execution of QMASM, QASM and Blackbird possible in web and mobile environments that don't have access to heavy compile toolchains, enabling adoption and scientific research. △ Less

Submitted 16 October, 2023; originally announced October 2023.

Pump, Dump, and then What? The Long-Term Impact of Cryptocurrency Pump-and-Dump Schemes

Authors: Joshua Clough, Matthew Edwards

Abstract: The pump and dump scheme is a form of market manipulation attack in which coordinated actors drive up the price of an asset in order to sell at a higher price. Due in part to a lack of enforcement, these schemes are widespread within the cryptocurrency marketplace, but the negative impact of these events on the coins they target is not yet fully understood. Drawing upon a novel dataset of pump eve… ▽ More The pump and dump scheme is a form of market manipulation attack in which coordinated actors drive up the price of an asset in order to sell at a higher price. Due in part to a lack of enforcement, these schemes are widespread within the cryptocurrency marketplace, but the negative impact of these events on the coins they target is not yet fully understood. Drawing upon a novel dataset of pump events extracted from Telegram channels, an order of magnitude larger than the nearest comparable dataset in the literature, we explore the differing tactics of pumping channels and the long-term impact of pump and dump schemes across 765 coins. We find that, despite a short-term positive impact in some cases, the long-term impact of pump and dump schemes on the targeted assets is negative, amounting to an average 30% relative drop in price a year after the pump event. △ Less

Submitted 12 September, 2023; originally announced September 2023.

Automatic Scam-Baiting Using ChatGPT

Authors: Piyush Bajaj, Matthew Edwards

Abstract: Automatic scam-baiting is an online fraud countermeasure that involves automated systems responding to online fraudsters in order to waste their time and deplete their resources, diverting attackers away from real potential victims. Previous work has demonstrated that text generation systems are capable of engaging with attackers as automatic scam-baiters, but the fluency and coherence of generate… ▽ More Automatic scam-baiting is an online fraud countermeasure that involves automated systems responding to online fraudsters in order to waste their time and deplete their resources, diverting attackers away from real potential victims. Previous work has demonstrated that text generation systems are capable of engaging with attackers as automatic scam-baiters, but the fluency and coherence of generated text may be a limit to the effectiveness of such systems. In this paper, we report on the results of a month-long experiment comparing the effectiveness of two ChatGPT-based automatic scam-baiters to a control measure. Within our results, with engagement from over 250 real email fraudsters, we find that ChatGPT-based scam-baiters show a marked increase in scammer response rate and conversation length relative to the control measure, outperforming previous approaches. We discuss the implications of these results and practical considerations for wider deployment of automatic scam-baiting. △ Less

Submitted 4 September, 2023; originally announced September 2023.

Comments: Proceedings of the 7th International Workshop on Applications of AI, Cyber Security and Economics Data Analytics (ACE-2023) (in press)

ADVISE: AI-accelerated Design of Evidence Synthesis for Global Development

Authors: Kristen M. Edwards, Binyang Song, Jaron Porciello, Mark Engelbert, Carolyn Huang, Faez Ahmed

Abstract: When designing evidence-based policies and programs, decision-makers must distill key information from a vast and rapidly growing literature base. Identifying relevant literature from raw search results is time and resource intensive, and is often done by manual screening. In this study, we develop an AI agent based on a bidirectional encoder representations from transformers (BERT) model and inco… ▽ More When designing evidence-based policies and programs, decision-makers must distill key information from a vast and rapidly growing literature base. Identifying relevant literature from raw search results is time and resource intensive, and is often done by manual screening. In this study, we develop an AI agent based on a bidirectional encoder representations from transformers (BERT) model and incorporate it into a human team designing an evidence synthesis product for global development. We explore the effectiveness of the human-AI hybrid team in accelerating the evidence synthesis process. To further improve team efficiency, we enhance the human-AI hybrid team through active learning (AL). Specifically, we explore different sampling strategies, including random sampling, least confidence (LC) sampling, and highest priority (HP) sampling, to study their influence on the collaborative screening process. Results show that incorporating the BERT-based AI agent into the human team can reduce the human screening effort by 68.5% compared to the case of no AI assistance and by 16.8% compared to the case of using a support vector machine (SVM)-based AI agent for identifying 80% of all relevant documents. When we apply the HP sampling strategy for AL, the human screening effort can be reduced even more: by 78.3% for identifying 80% of all relevant documents compared to no AI assistance. We apply the AL-enhanced human-AI hybrid teaming workflow in the design process of three evidence gap maps (EGMs) for USAID and find it to be highly effective. These findings demonstrate how AI can accelerate the development of evidence synthesis products and promote timely evidence-based decision making in global development in a human-AI hybrid teaming context. △ Less

Submitted 1 May, 2023; originally announced May 2023.

Comments: 14 pages, 11 figures, to be published in the proceedings of IDETC-CIE 2023

Quantum Honest Byzantine Agreement as a Distributed Quantum Algorithm

Authors: Marcus Edwards

Abstract: We suggest that the Quantum Honest Byzantine Agreement (QHBA) protocol [1] essentially reduces consensus to coincidence. The volume of coincidence is the parameter that drives a receiver to echo its input. A lack of coincidence results in no output from a receiver. This is a similar mechanism therefore to the learning mechanism in cognitive modular neural architectures like Haikonen's architecture… ▽ More We suggest that the Quantum Honest Byzantine Agreement (QHBA) protocol [1] essentially reduces consensus to coincidence. The volume of coincidence is the parameter that drives a receiver to echo its input. A lack of coincidence results in no output from a receiver. This is a similar mechanism therefore to the learning mechanism in cognitive modular neural architectures like Haikonen's architecture [2]. We introduce a simple feedback mechanism and quantum neuron to realize a hybrid quantum / classical machine learning network of simple nodes. △ Less

Submitted 15 January, 2023; v1 submitted 7 January, 2023; originally announced January 2023.

A Survey of Relevant Text Mining Technology

Authors: Claudia Peersman, Matthew Edwards, Emma Williams, Awais Rashid

Abstract: Recent advances in text mining and natural language processing technology have enabled researchers to detect an authors identity or demographic characteristics, such as age and gender, in several text genres by automatically analysing the variation of linguistic characteristics. However, applying such techniques in the wild, i.e., in both cybercriminal and regular online social media, differs from… ▽ More Recent advances in text mining and natural language processing technology have enabled researchers to detect an authors identity or demographic characteristics, such as age and gender, in several text genres by automatically analysing the variation of linguistic characteristics. However, applying such techniques in the wild, i.e., in both cybercriminal and regular online social media, differs from more general applications in that its defining characteristics are both domain and process dependent. This gives rise to a number of challenges of which contemporary research has only scratched the surface. More specifically, a text mining approach applied on social media communications typically has no control over the dataset size, the number of available communications will vary across users. Hence, the system has to be robust towards limited data availability. Additionally, the quality of the data cannot be guaranteed. As a result, the approach needs to be tolerant to a certain degree of linguistic noise (for example, abbreviations, non-standard language use, spelling variations and errors). Finally, in the context of cybercriminal fora, it has to be robust towards deceptive or adversarial behaviour, i.e. offenders who attempt to hide their criminal intentions (obfuscation) or who assume a false digital persona (imitation), potentially using coded language. In this work we present a comprehensive survey that discusses the problems that have already been addressed in current literature and review potential solutions. Additionally, we highlight which areas need to be given more attention. △ Less

Submitted 28 November, 2022; originally announced November 2022.

Active Countermeasures for Email Fraud

Authors: Wentao Chen, Fuzhou Wang, Matthew Edwards

Abstract: As a major component of online crime, email-based fraud is a threat that causes substantial economic losses every year. To counteract these scammers, volunteers called scam-baiters play the roles of victims, reply to scammers, and try to waste their time and attention with long and unproductive conversations. To curb email fraud and magnify the effectiveness of scam-baiting, we developed and deplo… ▽ More As a major component of online crime, email-based fraud is a threat that causes substantial economic losses every year. To counteract these scammers, volunteers called scam-baiters play the roles of victims, reply to scammers, and try to waste their time and attention with long and unproductive conversations. To curb email fraud and magnify the effectiveness of scam-baiting, we developed and deployed an expandable scam-baiting mailserver that can conduct scam-baiting activities automatically. We implemented three reply strategies using three different models and conducted a one-month-long experiment during which we elicited 150 messages from 130 different scammers. We compare the performance of each strategy at attracting and holding the attention of scammers, finding tradeoffs between human-written and automatically-generated response strategies. We also demonstrate that scammers can be engaged concurrently by multiple servers deploying these strategies in a second experiment, which used two server instances to contact 92 different scammers over 12 days. We release both our platform and a dataset containing conversations between our automatic scam-baiters and real human scammers, to support future work in preventing online fraud. △ Less

Submitted 1 June, 2023; v1 submitted 26 October, 2022; originally announced October 2022.

Journal ref: 2023 IEEE 8th European Symposium on Security and Privacy (EuroS&P)

Generating Diverse Indoor Furniture Arrangements

Authors: Ya-Chuan Hsu, Matthew C. Fontaine, Sam Earle, Maria Edwards, Julian Togelius, Stefanos Nikolaidis

Abstract: We present a method for generating arrangements of indoor furniture from human-designed furniture layout data. Our method creates arrangements that target specified diversity, such as the total price of all furniture in the room and the number of pieces placed. To generate realistic furniture arrangement, we train a generative adversarial network (GAN) on human-designed layouts. To target specific… ▽ More We present a method for generating arrangements of indoor furniture from human-designed furniture layout data. Our method creates arrangements that target specified diversity, such as the total price of all furniture in the room and the number of pieces placed. To generate realistic furniture arrangement, we train a generative adversarial network (GAN) on human-designed layouts. To target specific diversity in the arrangements, we optimize the latent space of the GAN via a quality diversity algorithm to generate a diverse arrangement collection. Experiments show our approach discovers a set of arrangements that are similar to human-designed layouts but varies in price and number of furniture pieces. △ Less

Submitted 20 June, 2022; originally announced June 2022.

Automatic User Profiling in Darknet Markets: a Scalability Study

Authors: Claudia Peersman, Matthew Edwards, Emma Williams, Awais Rashid

Abstract: In this study, we investigate the scalability of state-of-the-art user profiling technologies across different online domains. More specifically, this work aims to understand the reliability and limitations of current computational stylometry approaches when these are applied to underground fora in which user populations potentially differ from other online platforms (predominantly male, younger a… ▽ More In this study, we investigate the scalability of state-of-the-art user profiling technologies across different online domains. More specifically, this work aims to understand the reliability and limitations of current computational stylometry approaches when these are applied to underground fora in which user populations potentially differ from other online platforms (predominantly male, younger age and greater computer use) and cyber offenders who attempt to hide their identity. Because no ground truth is available and no validated criminal data from historic investigations is available for validation purposes, we have collected new data from clearweb forums that do include user demographics and could be more closely related to underground fora in terms of user population (e.g., tech communities) than commonly used social media benchmark datasets showing a more balanced user population. △ Less

Submitted 24 March, 2022; originally announced March 2022.

Understanding motivations and characteristics of financially-motivated cybercriminals

Authors: Claudia Peersman, Emma Williams, Matthew Edwards, Awais Rashid

Abstract: Background: Cyber offences, such as hacking, malware creation and distribution, and online fraud, present a substantial threat to organizations attempting to safeguard their data and information. By understanding the evolving characteristics and motivations of individuals involved in these activities, and the threats that they may pose, cyber security practitioners will be better placed to underst… ▽ More Background: Cyber offences, such as hacking, malware creation and distribution, and online fraud, present a substantial threat to organizations attempting to safeguard their data and information. By understanding the evolving characteristics and motivations of individuals involved in these activities, and the threats that they may pose, cyber security practitioners will be better placed to understand and assess current threats to their systems and the range of socio-technical mitigations that may best reduce these. Aim: The reported work-in-progress aims to explore the extent to which findings from prior academic literature regarding the characteristics and motivations of offenders engaging in financially-motivated, cyber-dependent crime are supported by the contemporary experiences and perspectives of practitioners currently working in the cyber crime field. Method: A targeted, online survey was developed consisting of both closed and open-ended questions relating to current cyber threats and the characteristics and motivations of offenders engaged in these activities. Sixteen practitioners working in law enforcement-related domains in the cyber crime field completed the survey, providing a combination of qualitative and quantitative data for analysis. △ Less

Submitted 28 March, 2022; v1 submitted 16 March, 2022; originally announced March 2022.

How darknet market users learned to worry more and love PGP: Analysis of security advice on darknet marketplaces

Authors: Andrew C. Dwyer, Joseph Hallett, Claudia Peersman, Matthew Edwards, Brittany I. Davidson, Awais Rashid

Abstract: Darknet marketplaces, accessible through, Tor are where users can buy illicit goods, and learn to hide from law enforcement. We surveyed the advice on these markets and found valid security advice mixed up with paranoid threat models and a reliance on privacy tools dismissed as unusable by the mainstream. Darknet marketplaces, accessible through, Tor are where users can buy illicit goods, and learn to hide from law enforcement. We surveyed the advice on these markets and found valid security advice mixed up with paranoid threat models and a reliance on privacy tools dismissed as unusable by the mainstream. △ Less

Submitted 16 March, 2022; originally announced March 2022.

Characterising Cybercriminals: A Review

Authors: Matthew Edwards, Emma Williams, Claudia Peersman, Awais Rashid

Abstract: This review provides an overview of current research on the known characteristics and motivations of offenders engaging in cyber-dependent crimes. Due to the shifting dynamics of cybercriminal behaviour, and the availability of prior reviews in 2013, this review focuses on original research conducted from 2012 onwards, although some older studies that were not included in prior reviews are also co… ▽ More This review provides an overview of current research on the known characteristics and motivations of offenders engaging in cyber-dependent crimes. Due to the shifting dynamics of cybercriminal behaviour, and the availability of prior reviews in 2013, this review focuses on original research conducted from 2012 onwards, although some older studies that were not included in prior reviews are also considered. As a basis for interpretation of results, a limited quality assessment was also carried out on included studies through examination of key indicators. △ Less

Submitted 15 February, 2022; originally announced February 2022.

Exploring open-ended gameplay features with Micro RollerCoaster Tycoon

Authors: Michael Cerny Green, Victoria Yen, Sam Earle, Dipika Rajesh, Maria Edwards, L. B. Soros

Abstract: This paper introduces MicroRCT, a novel open source simulator inspired by the theme park sandbox game RollerCoaster Tycoon. The goal in MicroRCT is to place rides and shops in an amusement park to maximize profit earned from park guests. Thus, the challenges for game AI include both selecting high-earning attractions and placing them in locations that are convenient to guests. In this paper, the M… ▽ More This paper introduces MicroRCT, a novel open source simulator inspired by the theme park sandbox game RollerCoaster Tycoon. The goal in MicroRCT is to place rides and shops in an amusement park to maximize profit earned from park guests. Thus, the challenges for game AI include both selecting high-earning attractions and placing them in locations that are convenient to guests. In this paper, the MAP-Elites algorithm is used to generate a diversity of park layouts, exploring two theoretical questions about evolutionary algorithms and game design: 1) Is there a benefit to starting from a minimal starting point for evolution and complexifying incrementally? and 2) What are the effects of resource limitations on creativity and optimization? Results indicate that building from scratch with no costs results in the widest diversity of high-performing designs. △ Less

Submitted 10 May, 2021; originally announced May 2021.

Comments: 8 pages, 10 figures, submitted to Foundations of Digital Games Conference 2021

Learning Controllable Content Generators

Authors: Sam Earle, Maria Edwards, Ahmed Khalifa, Philip Bontrager, Julian Togelius

Abstract: It has recently been shown that reinforcement learning can be used to train generators capable of producing high-quality game levels, with quality defined in terms of some user-specified heuristic. To ensure that these generators' output is sufficiently diverse (that is, not amounting to the reproduction of a single optimal level configuration), the generation process is constrained such that the… ▽ More It has recently been shown that reinforcement learning can be used to train generators capable of producing high-quality game levels, with quality defined in terms of some user-specified heuristic. To ensure that these generators' output is sufficiently diverse (that is, not amounting to the reproduction of a single optimal level configuration), the generation process is constrained such that the initial seed results in some variance in the generator's output. However, this results in a loss of control over the generated content for the human user. We propose to train generators capable of producing controllably diverse output, by making them "goal-aware." To this end, we add conditional inputs representing how close a generator is to some heuristic, and also modify the reward mechanism to incorporate that value. Testing on multiple domains, we show that the resulting level generators are capable of exploring the space of possible levels in a targeted, controllable manner, producing levels of comparable quality as their goal-unaware counterparts, that are diverse along designer-specified dimensions. △ Less

Submitted 6 May, 2021; originally announced May 2021.

Comments: 8 pages, 11 figures, submitted to CoG '21

The best laid plans or lack thereof: Security decision-making of different stakeholder groups

Authors: Benjamin Shreeve, Joseph Hallett, Matthew Edwards, Kopo M. Ramokapane, Richard Atkins, Awais Rashid

Abstract: Cyber security requirements are influenced by the priorities and decisions of a range of stakeholders. Board members and CISOs determine strategic priorities. Managers have responsibility for resource allocation and project management. Legal professionals concern themselves with regulatory compliance. Little is understood about how the security decision-making approaches of these different stakeho… ▽ More Cyber security requirements are influenced by the priorities and decisions of a range of stakeholders. Board members and CISOs determine strategic priorities. Managers have responsibility for resource allocation and project management. Legal professionals concern themselves with regulatory compliance. Little is understood about how the security decision-making approaches of these different stakeholders contrast, and if particular groups of stakeholders have a better appreciation of security requirements during decision-making. Are risk analysts better decision makers than CISOs? Do security experts exhibit more effective strategies than board members? This paper explores the effect that different experience and diversity of expertise has on the quality of a team's cyber security decision-making and whether teams with members from more varied backgrounds perform better than those with more focused, homogeneous skill sets. Using data from 208 sessions and 948 players of a tabletop game run in the wild by a major national organization over 16 months, we explore how choices are affected by player background (e.g.,~cyber security experts versus risk analysts, board-level decision makers versus technical experts) and different team make-ups (homogeneous teams of security experts versus various mixes). We find that no group of experts makes significantly better game decisions than anyone else, and that their biases lead them to not fully comprehend what they are defending or how the defenses work. △ Less

Submitted 1 April, 2021; originally announced April 2021.

Comments: 13 pages plus 2 page appendix. IEEE Transactions on Software Engineering 2020

MSC Class: 00Axx ACM Class: I.0

Achieving Multi-Port Memory Performance on Single-Port Memory with Coding Techniques

Authors: Hardik Jain, Matthew Edwards, Ethan Elenberg, Ankit Singh Rawat, Sriram Vishwanath

Abstract: Many performance critical systems today must rely on performance enhancements, such as multi-port memories, to keep up with the increasing demand of memory-access capacity. However, the large area footprints and complexity of existing multi-port memory designs limit their applicability. This paper explores a coding theoretic framework to address this problem. In particular, this paper introduces a… ▽ More Many performance critical systems today must rely on performance enhancements, such as multi-port memories, to keep up with the increasing demand of memory-access capacity. However, the large area footprints and complexity of existing multi-port memory designs limit their applicability. This paper explores a coding theoretic framework to address this problem. In particular, this paper introduces a framework to encode data across multiple single-port memory banks in order to {\em algorithmically} realize the functionality of multi-port memory. This paper proposes three code designs with significantly less storage overhead compared to the existing replication based emulations of multi-port memories. To further improve performance, we also demonstrate a memory controller design that utilizes redundancy across coded memory banks to more efficiently schedule read and write requests sent across multiple cores. Furthermore, guided by DRAM traces, the paper explores {\em dynamic coding} techniques to improve the efficiency of the coding based memory design. We then show significant performance improvements in critical word read and write latency in the proposed coded-memory design when compared to a traditional uncoded-memory design. △ Less

Submitted 27 January, 2020; originally announced January 2020.

Comments: 10 pages, 20 figures, ICICT 2020 conference

A Review of Quantum and Hybrid Quantum / Classical Blockchain Protocols

Authors: Marcus Edwards, Atefeh Mashatan, Shohini Ghose

Abstract: Blockchain technology is facing critical issues of scalability, efficiency and sustainability. These problems are necessary to solve if blockchain is to become a technology that can be used responsibly. Useful quantum computers could potentially be developed by the time that blockchain will be widely implemented for mission-critical work at financial and other institutions. Quantum computing will… ▽ More Blockchain technology is facing critical issues of scalability, efficiency and sustainability. These problems are necessary to solve if blockchain is to become a technology that can be used responsibly. Useful quantum computers could potentially be developed by the time that blockchain will be widely implemented for mission-critical work at financial and other institutions. Quantum computing will not only cause challenges for blockchain, but can also be harnessed to better implement parts of blockchain technologies including cryptocurrencies. We review the work that has been done in the area of quantum blockchain and hybrid quantum-classical blockchain technology and discuss open questions that remain. △ Less

Submitted 17 December, 2019; originally announced December 2019.

Comments: 27 pages, 1 figure

Journal ref: Quantum Inf Process 19, 184 (2020)

EigenRank by Committee: A Data Subset Selection and Failure Prediction paradigm for Robust Deep Learning based Medical Image Segmentation

Authors: Bilwaj Gaonkar, Joel Beckett, Mark Attiah, Christine Ahn, Matthew Edwards, Bayard Wilson, Azim Laiwalla, Banafsheh Salehi, Bryan Yoo, Alex Bui, Luke Macyszyn

Abstract: Translation of fully automated deep learning based medical image segmentation technologies to clinical workflows face two main algorithmic challenges. The first, is the collection and archival of large quantities of manually annotated ground truth data for both training and validation. The second is the relative inability of the majority of deep learning based segmentation techniques to alert phys… ▽ More Translation of fully automated deep learning based medical image segmentation technologies to clinical workflows face two main algorithmic challenges. The first, is the collection and archival of large quantities of manually annotated ground truth data for both training and validation. The second is the relative inability of the majority of deep learning based segmentation techniques to alert physicians to a likely segmentation failure. Here we propose a novel algorithm, named `Eigenrank' which addresses both of these challenges. Eigenrank can select for manual labeling, a subset of medical images from a large database, such that a U-Net trained on this subset is superior to one trained on a randomly selected subset of the same size. Eigenrank can also be used to pick out, cases in a large database, where deep learning segmentation will fail. We present our algorithm, followed by results and a discussion of how Eigenrank exploits the Von Neumann information to perform both data subset selection and failure prediction for medical image segmentation using deep learning. △ Less

Submitted 18 January, 2021; v1 submitted 17 August, 2019; originally announced August 2019.

MSC Class: 68T45 (Primary) 68T05; 68T20 (Secondary) ACM Class: I.5.4; I.4.6

Journal ref: Medical Image Analysis, Volume 67, 2021, Medical Image Analysis, Volume 67,2021,101834,ISSN 1361-8415,

Automatically Dismantling Online Dating Fraud

Authors: Guillermo Suarez-Tangil, Matthew Edwards, Claudia Peersman, Gianluca Stringhini, Awais Rashid, Monica Whitty

Abstract: Online romance scams are a prevalent form of mass-marketing fraud in the West, and yet few studies have addressed the technical or data-driven responses to this problem. In this type of scam, fraudsters craft fake profiles and manually interact with their victims. Because of the characteristics of this type of fraud and of how dating sites operate, traditional detection methods (e.g., those used i… ▽ More Online romance scams are a prevalent form of mass-marketing fraud in the West, and yet few studies have addressed the technical or data-driven responses to this problem. In this type of scam, fraudsters craft fake profiles and manually interact with their victims. Because of the characteristics of this type of fraud and of how dating sites operate, traditional detection methods (e.g., those used in spam filtering) are ineffective. In this paper, we present the results of a multi-pronged investigation into the archetype of online dating profiles used in this form of fraud, including their use of demographics, profile descriptions, and images, shedding light on both the strategies deployed by scammers to appeal to victims and the traits of victims themselves. Further, in response to the severe financial and psychological harm caused by dating fraud, we develop a system to detect romance scammers on online dating platforms. Our work presents the first system for automatically detecting this fraud. Our aim is to provide an early detection system to stop romance scammers as they create fraudulent profiles or before they engage with potential victims. Previous research has indicated that the victims of romance scams score highly on scales for idealized romantic beliefs. We combine a range of structured, unstructured, and deep-learned features that capture these beliefs. No prior work has fully analyzed whether these notions of romance introduce traits that could be leveraged to build a detection system. Our ensemble machine-learning approach is robust to the omission of profile details and performs at high accuracy (97\%). The system enables development of automated tools for dating site providers and individual users. △ Less

Submitted 30 May, 2019; v1 submitted 29 May, 2019; originally announced May 2019.

The one comparing narrative social network extraction techniques

Authors: Michelle Edwards, Lewis Mitchell, Jonathan Tuke, Matthew Roughan

Abstract: Analysing narratives through their social networks is an expanding field in quantitative literary studies. Manually extracting a social network from any narrative can be time consuming, so automatic extraction methods of varying complexity have been developed. However, the effect of different extraction methods on the analysis is unknown. Here we model and compare three extraction methods for soci… ▽ More Analysing narratives through their social networks is an expanding field in quantitative literary studies. Manually extracting a social network from any narrative can be time consuming, so automatic extraction methods of varying complexity have been developed. However, the effect of different extraction methods on the analysis is unknown. Here we model and compare three extraction methods for social networks in narratives: manual extraction, co-occurrence automated extraction and automated extraction using machine learning. Although the manual extraction method produces more precise results in the network analysis, it is much more time consuming and the automatic extraction methods yield comparable conclusions for density, centrality measures and edge weights. Our results provide evidence that social networks extracted automatically are reliable for many analyses. We also describe which aspects of analysis are not reliable with such a social network. We anticipate that our findings will make it easier to analyse more narratives, which help us improve our understanding of how stories are written and evolve, and how people interact with each other. △ Less

Submitted 4 November, 2018; originally announced November 2018.

Extreme Augmentation : Can deep learning based medical image segmentation be trained using a single manually delineated scan?

Authors: Bilwaj Gaonkar, Matthew Edwards, Alex Bui, Matthew Brown, Luke Macyszyn

Abstract: Yes, it can. Data augmentation is perhaps the oldest preprocessing step in computer vision literature. Almost every computer vision model trained on imaging data uses some form of augmentation. In this paper, we use the inter-vertebral disk segmentation task alongside a deep residual U-Net as the learning model, to explore the effectiveness of augmentation. In the extreme, we observed that a model… ▽ More Yes, it can. Data augmentation is perhaps the oldest preprocessing step in computer vision literature. Almost every computer vision model trained on imaging data uses some form of augmentation. In this paper, we use the inter-vertebral disk segmentation task alongside a deep residual U-Net as the learning model, to explore the effectiveness of augmentation. In the extreme, we observed that a model trained on patches extracted from just one scan, with each patch augmented 50 times; achieved a Dice score of 0.73 in a validation set of 40 cases. Qualitative evaluation indicated a clinically usable segmentation algorithm, which appropriately segments regions of interest, alongside limited false positive specks. When the initial patches are extracted from nine scans the average Dice coefficient jumps to 0.86 and most of the false positives disappear. While this still falls short of state-of-the-art deep learning based segmentation of discs reported in literature, qualitative examination reveals that it does yield segmentation, which can be amended by expert clinicians with minimal effort to generate additional data for training improved deep models. Extreme augmentation of training data, should thus be construed as a strategy for training deep learning based algorithms, when very little manually annotated data is available to work with. Models trained with extreme augmentation can then be used to accelerate the generation of manually labelled data. Hence, we show that extreme augmentation can be a valuable tool in addressing scaling up small imaging data sets to address medical image segmentation tasks. △ Less

Submitted 6 September, 2019; v1 submitted 3 October, 2018; originally announced October 2018.

The Interhospital Transfer Network for Very Low Birth Weight Infants in the United States

Authors: Munik Shrestha, Samuel V. Scarpino, Erika M. Edwards, Lucy T. Greenberg, Jeffrey D. Horbar

Abstract: Very low birth weight (VLBW) infants require specialized care in neonatal intensive care units. In the United States (U.S.), such infants frequently are transferred between hospitals. Although these neonatal transfer networks are important, both economically and for infant morbidity and mortality, the national-level pattern of neonatal transfers is largely unknown. Using data from Vermont Oxford N… ▽ More Very low birth weight (VLBW) infants require specialized care in neonatal intensive care units. In the United States (U.S.), such infants frequently are transferred between hospitals. Although these neonatal transfer networks are important, both economically and for infant morbidity and mortality, the national-level pattern of neonatal transfers is largely unknown. Using data from Vermont Oxford Network on 44,753 births, 2,122 hospitals, and 9,722 inter-hospital infant transfers from 2015, we performed the largest analysis to date on the inter-hospital transfer network for VLBW infants in the U.S. We find that transfers are organized around regional communities, but that despite being largely within state boundaries, most communities often contain at least two hospitals in different states. To classify the structural variation in transfer pattern amongst these communities, we applied a spectral measure for regionalization and found an association between a community's degree of regionalization and their infant transfer rate, which was not utilized in detecting communities. We also demonstrate that the established measures of network centrality and hierarchy, e.g., the community-wide entropy in PageRank or betweenness centrality and number of distinct `layers' within a community, correlate weakly with our regionalization index and were not significantly associated with metrics on infant transfer rate. Our results suggest that the regionalization index captures novel information about the structural properties of VLBW infant transfer networks, have the practical implication of characterizing neonatal care in the U.S., and may apply more broadly to the role of centralizing forces in organizing complex adaptive systems. △ Less

Submitted 2 July, 2018; v1 submitted 26 January, 2018; originally announced February 2018.

Graph Based Convolutional Neural Network

Authors: Michael Edwards, Xianghua Xie

Abstract: The benefit of localized features within the regular domain has given rise to the use of Convolutional Neural Networks (CNNs) in machine learning, with great proficiency in the image classification. The use of CNNs becomes problematic within the irregular spatial domain due to design and convolution of a kernel filter being non-trivial. One solution to this problem is to utilize graph signal proce… ▽ More The benefit of localized features within the regular domain has given rise to the use of Convolutional Neural Networks (CNNs) in machine learning, with great proficiency in the image classification. The use of CNNs becomes problematic within the irregular spatial domain due to design and convolution of a kernel filter being non-trivial. One solution to this problem is to utilize graph signal processing techniques and the convolution theorem to perform convolutions on the graph of the irregular domain to obtain feature map responses to learnt filters. We propose graph convolution and pooling operators analogous to those in the regular domain. We also provide gradient calculations on the input data and spectral filters, which allow for the deep learning of an irregular spatial domain problem. Signal filters take the form of spectral multipliers, applying convolution in the graph spectral domain. Applying smooth multipliers results in localized convolutions in the spatial domain, with smoother multipliers providing sharper feature maps. Algebraic Multigrid is presented as a graph pooling method, reducing the resolution of the graph through agglomeration of nodes between layers of the network. Evaluation of performance on the MNIST digit classification problem in both the regular and irregular domain is presented, with comparison drawn to standard CNN. The proposed graph CNN provides a deep learning method for the irregular domains present in the machine learning community, obtaining 94.23% on the regular grid, and 94.96% on a spatially irregular subsampled MNIST. △ Less

Submitted 28 September, 2016; originally announced September 2016.

Comments: 11 pages, accepted into BMVC 2016

From Pose to Activity: Surveying Datasets and Introducing CONVERSE

Authors: Michael Edwards, Jingjing Deng, Xianghua Xie

Abstract: We present a review on the current state of publicly available datasets within the human action recognition community; highlighting the revival of pose based methods and recent progress of understanding person-person interaction modeling. We categorize datasets regarding several key properties for usage as a benchmark dataset; including the number of class labels, ground truths provided, and appli… ▽ More We present a review on the current state of publicly available datasets within the human action recognition community; highlighting the revival of pose based methods and recent progress of understanding person-person interaction modeling. We categorize datasets regarding several key properties for usage as a benchmark dataset; including the number of class labels, ground truths provided, and application domain they occupy. We also consider the level of abstraction of each dataset; grouping those that present actions, interactions and higher level semantic activities. The survey identifies key appearance and pose based datasets, noting a tendency for simplistic, emphasized, or scripted action classes that are often readily definable by a stable collection of sub-action gestures. There is a clear lack of datasets that provide closely related actions, those that are not implicitly identified via a series of poses and gestures, but rather a dynamic set of interactions. We therefore propose a novel dataset that represents complex conversational interactions between two individuals via 3D pose. 8 pairwise interactions describing 7 separate conversation based scenarios were collected using two Kinect depth sensors. The intention is to provide events that are constructed from numerous primitive actions, interactions and motions, over a period of time; providing a set of subtle action classes that are more representative of the real world, and a challenge to currently developed recognition methodologies. We believe this is among one of the first datasets devoted to conversational interaction classification using 3D pose features and the attributed papers show this task is indeed possible. The full dataset is made publicly available to the research community at www.csvision.swansea.ac.uk/converse. △ Less

Submitted 19 November, 2015; v1 submitted 18 November, 2015; originally announced November 2015.

Comments: Presentation of pose-based conversational human interaction dataset, review of current appearance and depth based action recognition datasets, public dataset, 38 pages