Showing 1–16 of 16 results for author: Mann, G

Improving the Region of Attraction of a Multi-rotor UAV by Estimating Unknown Disturbances

Authors: Sachithra Atapattu, Oscar De Silva, Thumeera R Wanasinghe, George K I Mann, Raymond G Gosine

Abstract: This study presents a machine learning-aided approach to accurately estimate the region of attraction (ROA) of a multi-rotor unmanned aerial vehicle (UAV) controlled using a linear quadratic regulator (LQR) controller. Conventional ROA estimation approaches rely on a nominal dynamic model for ROA calculation, leading to inaccurate estimation due to unknown dynamics and disturbances associated with… ▽ More This study presents a machine learning-aided approach to accurately estimate the region of attraction (ROA) of a multi-rotor unmanned aerial vehicle (UAV) controlled using a linear quadratic regulator (LQR) controller. Conventional ROA estimation approaches rely on a nominal dynamic model for ROA calculation, leading to inaccurate estimation due to unknown dynamics and disturbances associated with the physical system. To address this issue, our study utilizes a neural network to predict these unknown disturbances of a planar quadrotor. The nominal model integrated with the learned disturbances is then employed to calculate the ROA of the planer quadrotor using a graphical technique. The estimated ROA is then compared with the ROA calculated using Lyapunov analysis and the graphical approach without incorporating the learned disturbances. The results illustrated that the proposed method provides a more accurate estimation of the ROA, while the conventional Lyapunov-based estimation tends to be more conservative. △ Less

Submitted 30 August, 2024; originally announced September 2024.

A Generative Adversarial Network-based Method for LiDAR-Assisted Radar Image Enhancement

Authors: Thakshila Thilakanayake, Oscar De Silva, Thumeera R. Wanasinghe, George K. Mann, Awantha Jayasiri

Abstract: This paper presents a generative adversarial network (GAN) based approach for radar image enhancement. Although radar sensors remain robust for operations under adverse weather conditions, their application in autonomous vehicles (AVs) is commonly limited by the low-resolution data they produce. The primary goal of this study is to enhance the radar images to better depict the details and features… ▽ More This paper presents a generative adversarial network (GAN) based approach for radar image enhancement. Although radar sensors remain robust for operations under adverse weather conditions, their application in autonomous vehicles (AVs) is commonly limited by the low-resolution data they produce. The primary goal of this study is to enhance the radar images to better depict the details and features of the environment, thereby facilitating more accurate object identification in AVs. The proposed method utilizes high-resolution, two-dimensional (2D) projected light detection and ranging (LiDAR) point clouds as ground truth images and low-resolution radar images as inputs to train the GAN. The ground truth images were obtained through two main steps. First, a LiDAR point cloud map was generated by accumulating raw LiDAR scans. Then, a customized LiDAR point cloud cropping and projection method was employed to obtain 2D projected LiDAR point clouds. The inference process of the proposed method relies solely on radar images to generate an enhanced version of them. The effectiveness of the proposed method is demonstrated through both qualitative and quantitative results. These results show that the proposed method can generate enhanced images with clearer object representation compared to the input radar images, even under adverse weather conditions. △ Less

Submitted 30 August, 2024; originally announced September 2024.

Scaling Laws for Galaxy Images

Authors: Mike Walmsley, Micah Bowles, Anna M. M. Scaife, Jason Shingirai Makechemu, Alexander J. Gordon, Annette M. N. Ferguson, Robert G. Mann, James Pearson, Jürgen J. Popp, Jo Bovy, Josh Speagle, Hugh Dickinson, Lucy Fortson, Tobias Géron, Sandor Kruk, Chris J. Lintott, Kameswara Mantha, Devina Mohan, David O'Ryan, Inigo V. Slijepevic

Abstract: We present the first systematic investigation of supervised scaling laws outside of an ImageNet-like context - on images of galaxies. We use 840k galaxy images and over 100M annotations by Galaxy Zoo volunteers, comparable in scale to Imagenet-1K. We find that adding annotated galaxy images provides a power law improvement in performance across all architectures and all tasks, while adding trainab… ▽ More We present the first systematic investigation of supervised scaling laws outside of an ImageNet-like context - on images of galaxies. We use 840k galaxy images and over 100M annotations by Galaxy Zoo volunteers, comparable in scale to Imagenet-1K. We find that adding annotated galaxy images provides a power law improvement in performance across all architectures and all tasks, while adding trainable parameters is effective only for some (typically more subjectively challenging) tasks. We then compare the downstream performance of finetuned models pretrained on either ImageNet-12k alone vs. additionally pretrained on our galaxy images. We achieve an average relative error rate reduction of 31% across 5 downstream tasks of scientific interest. Our finetuned models are more label-efficient and, unlike their ImageNet-12k-pretrained equivalents, often achieve linear transfer performance equal to that of end-to-end finetuning. We find relatively modest additional downstream benefits from scaling model size, implying that scaling alone is not sufficient to address our domain gap, and suggest that practitioners with qualitatively different images might benefit more from in-domain adaption followed by targeted downstream labelling. △ Less

Submitted 3 April, 2024; originally announced April 2024.

Comments: 10+6 pages, 12 figures. Appendix C2 based on arxiv:2206.11927. Code, demos, documentation at https://github.com/mwalmsley/zoobot

MUN-FRL: A Visual Inertial LiDAR Dataset for Aerial Autonomous Navigation and Mapping

Authors: Ravindu G. Thalagala, Sahan M. Gunawardena, Oscar De Silva, Awantha Jayasiri, Arthur Gubbels, George K. I Mann, Raymond G. Gosine

Abstract: This paper presents a unique outdoor aerial visual-inertial-LiDAR dataset captured using a multi-sensor payload to promote the global navigation satellite system (GNSS)-denied navigation research. The dataset features flight distances ranging from 300m to 5km, collected using a DJI M600 hexacopter drone and the National Research Council (NRC) Bell 412 Advanced Systems Research Aircraft (ASRA). The… ▽ More This paper presents a unique outdoor aerial visual-inertial-LiDAR dataset captured using a multi-sensor payload to promote the global navigation satellite system (GNSS)-denied navigation research. The dataset features flight distances ranging from 300m to 5km, collected using a DJI M600 hexacopter drone and the National Research Council (NRC) Bell 412 Advanced Systems Research Aircraft (ASRA). The dataset consists of hardware synchronized monocular images, IMU measurements, 3D LiDAR point-clouds, and high-precision real-time kinematic (RTK)-GNSS based ground truth. Ten datasets were collected as ROS bags over 100 mins of outdoor environment footage ranging from urban areas, highways, hillsides, prairies, and waterfronts. The datasets were collected to facilitate the development of visual-inertial-LiDAR odometry and mapping algorithms, visual-inertial navigation algorithms, object detection, segmentation, and landing zone detection algorithms based upon real-world drone and full-scale helicopter data. All the datasets contain raw sensor measurements, hardware timestamps, and spatio-temporally aligned ground truth. The intrinsic and extrinsic calibrations of the sensors are also provided along with raw calibration datasets. A performance summary of state-of-the-art methods applied on the datasets is also provided. △ Less

Submitted 12 October, 2023; originally announced October 2023.

BloombergGPT: A Large Language Model for Finance

Authors: Shijie Wu, Ozan Irsoy, Steven Lu, Vadim Dabravolski, Mark Dredze, Sebastian Gehrmann, Prabhanjan Kambadur, David Rosenberg, Gideon Mann

Abstract: The use of NLP in the realm of financial technology is broad and complex, with applications ranging from sentiment analysis and named entity recognition to question answering. Large Language Models (LLMs) have been shown to be effective on a variety of tasks; however, no LLM specialized for the financial domain has been reported in literature. In this work, we present BloombergGPT, a 50 billion pa… ▽ More The use of NLP in the realm of financial technology is broad and complex, with applications ranging from sentiment analysis and named entity recognition to question answering. Large Language Models (LLMs) have been shown to be effective on a variety of tasks; however, no LLM specialized for the financial domain has been reported in literature. In this work, we present BloombergGPT, a 50 billion parameter language model that is trained on a wide range of financial data. We construct a 363 billion token dataset based on Bloomberg's extensive data sources, perhaps the largest domain-specific dataset yet, augmented with 345 billion tokens from general purpose datasets. We validate BloombergGPT on standard LLM benchmarks, open financial benchmarks, and a suite of internal benchmarks that most accurately reflect our intended usage. Our mixed dataset training leads to a model that outperforms existing models on financial tasks by significant margins without sacrificing performance on general LLM benchmarks. Additionally, we explain our modeling choices, training process, and evaluation methodology. We release Training Chronicles (Appendix C) detailing our experience in training BloombergGPT. △ Less

Submitted 21 December, 2023; v1 submitted 30 March, 2023; originally announced March 2023.

Comments: Updated to include Training Chronicles (Appendix C)

Dual Reinforcement-Based Specification Generation for Image De-Rendering

Authors: Ramakanth Pasunuru, David Rosenberg, Gideon Mann, Mohit Bansal

Abstract: Advances in deep learning have led to promising progress in inferring graphics programs by de-rendering computer-generated images. However, current methods do not explore which decoding methods lead to better inductive bias for inferring graphics programs. In our work, we first explore the effectiveness of LSTM-RNN versus Transformer networks as decoders for order-independent graphics programs. Si… ▽ More Advances in deep learning have led to promising progress in inferring graphics programs by de-rendering computer-generated images. However, current methods do not explore which decoding methods lead to better inductive bias for inferring graphics programs. In our work, we first explore the effectiveness of LSTM-RNN versus Transformer networks as decoders for order-independent graphics programs. Since these are sequence models, we must choose an ordering of the objects in the graphics programs for likelihood training. We found that the LSTM performance was highly sensitive to the sequence ordering (random order vs. pattern-based order), while Transformer performance was roughly independent of the sequence ordering. Further, we present a policy gradient based reinforcement learning approach for better inductive bias in the decoder via multiple diverse rewards based both on the graphics program specification and the rendered image. We also explore the combination of these complementary rewards. We achieve state-of-the-art results on two graphics program generation datasets. △ Less

Submitted 2 March, 2021; originally announced March 2021.

Comments: AAAI 2021 Scientific Document Understanding Workshop (9 pages)

Improving Grey-Box Fuzzing by Modeling Program Behavior

Authors: Siddharth Karamcheti, Gideon Mann, David Rosenberg

Abstract: Grey-box fuzzers such as American Fuzzy Lop (AFL) are popular tools for finding bugs and potential vulnerabilities in programs. While these fuzzers have been able to find vulnerabilities in many widely used programs, they are not efficient; of the millions of inputs executed by AFL in a typical fuzzing run, only a handful discover unseen behavior or trigger a crash. The remaining inputs are redund… ▽ More Grey-box fuzzers such as American Fuzzy Lop (AFL) are popular tools for finding bugs and potential vulnerabilities in programs. While these fuzzers have been able to find vulnerabilities in many widely used programs, they are not efficient; of the millions of inputs executed by AFL in a typical fuzzing run, only a handful discover unseen behavior or trigger a crash. The remaining inputs are redundant, exhibiting behavior that has already been observed. Here, we present an approach to increase the efficiency of fuzzers like AFL by applying machine learning to directly model how programs behave. We learn a forward prediction model that maps program inputs to execution traces, training on the thousands of inputs collected during standard fuzzing. This learned model guides exploration by focusing on fuzzing inputs on which our model is the most uncertain (measured via the entropy of the predicted execution trace distribution). By focusing on executing inputs our learned model is unsure about, and ignoring any input whose behavior our model is certain about, we show that we can significantly limit wasteful execution. Through testing our approach on a set of binaries released as part of the DARPA Cyber Grand Challenge, we show that our approach is able to find a set of inputs that result in more code coverage and discovered crashes than baseline fuzzers with significantly fewer executions. △ Less

Submitted 21 November, 2018; originally announced November 2018.

Comments: 5 pages, 3 figures

Adaptive Grey-Box Fuzz-Testing with Thompson Sampling

Authors: Siddharth Karamcheti, Gideon Mann, David Rosenberg

Abstract: Fuzz testing, or "fuzzing," refers to a widely deployed class of techniques for testing programs by generating a set of inputs for the express purpose of finding bugs and identifying security flaws. Grey-box fuzzing, the most popular fuzzing strategy, combines light program instrumentation with a data driven process to generate new program inputs. In this work, we present a machine learning approa… ▽ More Fuzz testing, or "fuzzing," refers to a widely deployed class of techniques for testing programs by generating a set of inputs for the express purpose of finding bugs and identifying security flaws. Grey-box fuzzing, the most popular fuzzing strategy, combines light program instrumentation with a data driven process to generate new program inputs. In this work, we present a machine learning approach that builds on AFL, the preeminent grey-box fuzzer, by adaptively learning a probability distribution over its mutation operators on a program-specific basis. These operators, which are selected uniformly at random in AFL and mutational fuzzers in general, dictate how new inputs are generated, a core part of the fuzzer's efficacy. Our main contributions are two-fold: First, we show that a sampling distribution over mutation operators estimated from training programs can significantly improve performance of AFL. Second, we introduce a Thompson Sampling, bandit-based optimization approach that fine-tunes the mutator distribution adaptively, during the course of fuzzing an individual program. A set of experiments across complex programs demonstrates that tuning the mutational operator distribution generates sets of inputs that yield significantly higher code coverage and finds more crashes faster and more reliably than both baseline versions of AFL as well as other AFL-based learning approaches. △ Less

Submitted 24 August, 2018; originally announced August 2018.

Comments: Published as a workshop paper in the 11th ACM Workshop on Artificial Intelligence and Security (AISec '18) with the 25th ACM Conference on Computer and Communications Security (CCS '18)

Use of Docker for deployment and testing of astronomy software

Authors: D. Morris, S. Voutsinas, N. C. Hambly, R. G. Mann

Abstract: We describe preliminary investigations of using Docker for the deployment and testing of astronomy software. Docker is a relatively new containerisation technology that is developing rapidly and being adopted across a range of domains. It is based upon virtualization at operating system level, which presents many advantages in comparison to the more traditional hardware virtualization that underpi… ▽ More We describe preliminary investigations of using Docker for the deployment and testing of astronomy software. Docker is a relatively new containerisation technology that is developing rapidly and being adopted across a range of domains. It is based upon virtualization at operating system level, which presents many advantages in comparison to the more traditional hardware virtualization that underpins most cloud computing infrastructure today. A particular strength of Docker is its simple format for describing and managing software containers, which has benefits for software developers, system administrators and end users. We report on our experiences from two projects -- a simple activity to demonstrate how Docker works, and a more elaborate set of services that demonstrates more of its capabilities and what they can achieve within an astronomical context -- and include an account of how we solved problems through interaction with Docker's very active open source development community, which is currently the key to the most effective use of this rapidly-changing technology. △ Less

Submitted 11 July, 2017; originally announced July 2017.

Comments: 29 pages, 9 figures, accepted for publication in Astronomy and Computing, ref ASCOM199

The Right Invariant Nonlinear Complementary Filter for Low Cost Attitude and Heading Estimation of Platforms

Authors: Oscar De Silva, George K. I. Mann, Raymond G. Gosine

Abstract: This paper presents a novel filter with low computational demand to address the problem of orientation estimation of a robotic platform. This is conventionally addressed by extended Kalman filtering of measurements from a sensor suit which mainly includes accelerometers, gyroscopes, and a digital compass. Low cost robotic platforms demand simpler and computationally more efficient methods to addre… ▽ More This paper presents a novel filter with low computational demand to address the problem of orientation estimation of a robotic platform. This is conventionally addressed by extended Kalman filtering of measurements from a sensor suit which mainly includes accelerometers, gyroscopes, and a digital compass. Low cost robotic platforms demand simpler and computationally more efficient methods to address this filtering problem. Hence nonlinear observers with constant gains have emerged to assume this role. The nonlinear complementary filter is a popular choice in this domain which does not require covariance matrix propagation and associated computational overhead in its filtering algorithm. However, the gain tuning procedure of the complementary filter is not optimal, where it is often hand picked by trial and error. This process is counter intuitive to system noise based tuning capability offered by a stochastic filter like the Kalman filter. This paper proposes the right invariant formulation of the complementary filter, which preserves Kalman like system noise based gain tuning capability for the filter. The resulting filter exhibits efficient operation in elementary embedded hardware, intuitive system noise based gain tuning capability and accurate attitude estimation. The performance of the filter is validated using numerical simulations and by experimentally implementing the filter on an ARDrone 2.0 micro aerial vehicle platform. △ Less

Submitted 30 November, 2016; v1 submitted 30 November, 2016; originally announced November 2016.

Twitter as a Source of Global Mobility Patterns for Social Good

Authors: Mark Dredze, Manuel García-Herranz, Alex Rutherford, Gideon Mann

Abstract: Data on human spatial distribution and movement is essential for understanding and analyzing social systems. However existing sources for this data are lacking in various ways; difficult to access, biased, have poor geographical or temporal resolution, or are significantly delayed. In this paper, we describe how geolocation data from Twitter can be used to estimate global mobility patterns and add… ▽ More Data on human spatial distribution and movement is essential for understanding and analyzing social systems. However existing sources for this data are lacking in various ways; difficult to access, biased, have poor geographical or temporal resolution, or are significantly delayed. In this paper, we describe how geolocation data from Twitter can be used to estimate global mobility patterns and address these shortcomings. These findings will inform how this novel data source can be harnessed to address humanitarian and development efforts. △ Less

Submitted 20 June, 2016; originally announced June 2016.

Comments: Presented at 2016 ICML Workshop on #Data4Good: Machine Learning in Social Good Applications, New York, NY

Renewal Strings for Cleaning Astronomical Databases

Authors: Amos J. Storkey, Nigel C. Hambly, Christopher K. I. Williams, Robert G. Mann

Abstract: Large astronomical databases obtained from sky surveys such as the SuperCOSMOS Sky Surveys (SSS) invariably suffer from a small number of spurious records coming from artefactual effects of the telescope, satellites and junk objects in orbit around earth and physical defects on the photographic plate or CCD. Though relatively small in number these spurious records present a significant problem in… ▽ More Large astronomical databases obtained from sky surveys such as the SuperCOSMOS Sky Surveys (SSS) invariably suffer from a small number of spurious records coming from artefactual effects of the telescope, satellites and junk objects in orbit around earth and physical defects on the photographic plate or CCD. Though relatively small in number these spurious records present a significant problem in many situations where they can become a large proportion of the records potentially of interest to a given astronomer. In this paper we focus on the four most common causes of unwanted records in the SSS: satellite or aeroplane tracks, scratches fibres and other linear phenomena introduced to the plate, circular halos around bright stars due to internal reflections within the telescope and diffraction spikes near to bright stars. Accurate and robust techniques are needed for locating and flagging such spurious objects. We have developed renewal strings, a probabilistic technique combining the Hough transform, renewal processes and hidden Markov models which have proven highly effective in this context. The methods are applied to the SSS data to develop a dataset of spurious object detections, along with confidence measures, which can allow this unwanted data to be removed from consideration. These methods are general and can be adapted to any future astronomical survey data. △ Less

Submitted 7 August, 2014; originally announced August 2014.

Comments: Appears in Proceedings of the Nineteenth Conference on Uncertainty in Artificial Intelligence (UAI2003)

Report number: UAI-P-2003-PG-559-566

Astronomy and Computing: a New Journal for the Astronomical Computing Community

Authors: Alberto Accomazzi, Tamás Budavári, Christopher Fluke, Norman Gray, Robert G Mann, William O'Mullane, Andreas Wicenec, Michael Wise

Abstract: We introduce \emph{Astronomy and Computing}, a new journal for the growing population of people working in the domain where astronomy overlaps with computer science and information technology. The journal aims to provide a new communication channel within that community, which is not well served by current journals, and to help secure recognition of its true importance within modern astronomy. In… ▽ More We introduce \emph{Astronomy and Computing}, a new journal for the growing population of people working in the domain where astronomy overlaps with computer science and information technology. The journal aims to provide a new communication channel within that community, which is not well served by current journals, and to help secure recognition of its true importance within modern astronomy. In this inaugural editorial, we describe the rationale for creating the journal, outline its scope and ambitions, and seek input from the community in defining in detail how the journal should work towards its high-level goals. △ Less

Submitted 30 October, 2012; originally announced October 2012.

Comments: 5 pages, no figures; editorial for first edition of journal

AstroDAbis: Annotations and Cross-Matches for Remote Catalogues

Authors: Norman Gray, Robert G Mann, Dave Morris, Mark Holliman, Keith Noddle

Abstract: Astronomers are good at sharing data, but poorer at sharing knowledge. Almost all astronomical data ends up in open archives, and access to these is being simplified by the development of the global Virtual Observatory (VO). This is a great advance, but the fundamental problem remains that these archives contain only basic observational data, whereas all the astrophysical interpretation of that… ▽ More Astronomers are good at sharing data, but poorer at sharing knowledge. Almost all astronomical data ends up in open archives, and access to these is being simplified by the development of the global Virtual Observatory (VO). This is a great advance, but the fundamental problem remains that these archives contain only basic observational data, whereas all the astrophysical interpretation of that data -- which source is a quasar, which a low-mass star, and which an image artefact -- is contained in journal papers, with very little linkage back from the literature to the original data archives. It is therefore currently impossible for an astronomer to pose a query like "give me all sources in this data archive that have been identified as quasars" and this limits the effective exploitation of these archives, as the user of an archive has no direct means of taking advantage of the knowledge derived by its previous users. The AstroDAbis service aims to address this, in a prototype service enabling astronomers to record annotations and cross-identifications in the AstroDAbis service, annotating objects in other catalogues. We have deployed two interfaces to the annotations, namely one astronomy-specific one using the TAP protocol}, and a second exploiting generic Linked Open Data (LOD) and RDF techniques. △ Less

Submitted 25 November, 2011; originally announced November 2011.

Comments: 4 pages, 1 figure, to appear in Proceedings of ADASS XXI, Paris, 2011

ACM Class: J.2; H.2.8; H.5.4

Collaborative Astronomical Image Mosaics

Authors: Daniel S. Katz, G. Bruce Berriman, Robert G. Mann

Abstract: This chapter describes how astronomical imaging survey data have become a vital part of modern astronomy, how these data are archived and then served to the astronomical community through on-line data access portals. The Virtual Observatory, now under development, aims to make all these data accessible through a uniform set of interfaces. This chapter also describes the scientific need for one com… ▽ More This chapter describes how astronomical imaging survey data have become a vital part of modern astronomy, how these data are archived and then served to the astronomical community through on-line data access portals. The Virtual Observatory, now under development, aims to make all these data accessible through a uniform set of interfaces. This chapter also describes the scientific need for one common image processing task, that of composing individual images into large scale mosaics and introduces Montage as a tool for this task. Montage, as distributed, can be used in four ways: as a single thread/process on a single CPU, in parallel using MPI to distribute similar tasks across a parallel computer, in parallel using grid tools (Pegasus/DAGMan) to distributed tasks across a grid, or in parallel using a script-driven approach (Swift). An on-request web based Montage service is available for users who do not need to build a local version. We also introduce some work on a new scripted version of Montage, which offers ease of customization for users. Then, we discuss various ideas where Web 2.0 technologies can help the Montage community. △ Less

Submitted 23 November, 2010; originally announced November 2010.

Comments: 16 pages, 3 figures. To be published in "Reshaping Research and Development using Web 2.0-based technologies." Mark Baker, ed. Nova Science Publishers, Inc.(2011)

Looking Under the Hood : Tools for Diagnosing your Question Answering Engine

Authors: Eric Breck, Marc Light, Gideon S. Mann, Ellen Riloff, Brianne Brown Pranav Anand, Mats Rooth, Michael Thelen

Abstract: In this paper we analyze two question answering tasks : the TREC-8 question answering task and a set of reading comprehension exams. First, we show that Q/A systems perform better when there are multiple answer opportunities per question. Next, we analyze common approaches to two subproblems: term overlap for answer sentence identification, and answer typing for short answer extraction. We prese… ▽ More In this paper we analyze two question answering tasks : the TREC-8 question answering task and a set of reading comprehension exams. First, we show that Q/A systems perform better when there are multiple answer opportunities per question. Next, we analyze common approaches to two subproblems: term overlap for answer sentence identification, and answer typing for short answer extraction. We present general tools for analyzing the strengths and limitations of techniques for these subproblems. Our results quantify the limitations of both term overlap and answer typing to distinguish between competing answer candidates. △ Less

Submitted 3 July, 2001; originally announced July 2001.

Comments: Revision of paper appearing in the Proceedings of the Workshop on Open-Domain Question Answering

ACM Class: I.2.7