-
An Iterative Feedback Mechanism for Improving Natural Language Class Descriptions in Open-Vocabulary Object Detection
Authors:
Louis Y. Kim,
Michelle Karker,
Victoria Valledor,
Seiyoung C. Lee,
Karl F. Brzoska,
Margaret Duff,
Anthony Palladino
Abstract:
Recent advances in open-vocabulary object detection models will enable Automatic Target Recognition systems to be sustainable and repurposed by non-technical end-users for a variety of applications or missions. New, and potentially nuanced, classes can be defined with natural language text descriptions in the field, immediately before runtime, without needing to retrain the model. We present an ap…
▽ More
Recent advances in open-vocabulary object detection models will enable Automatic Target Recognition systems to be sustainable and repurposed by non-technical end-users for a variety of applications or missions. New, and potentially nuanced, classes can be defined with natural language text descriptions in the field, immediately before runtime, without needing to retrain the model. We present an approach for improving non-technical users' natural language text descriptions of their desired targets of interest, using a combination of analysis techniques on the text embeddings, and proper combinations of embeddings for contrastive examples. We quantify the improvement that our feedback mechanism provides by demonstrating performance with multiple publicly-available open-vocabulary object detection models.
△ Less
Submitted 21 March, 2025;
originally announced March 2025.
-
Comprehensive peak-width analysis in matter-wave diffraction under grazing incidence conditions
Authors:
Lee Yeong Kim,
Do Won Kang,
Jong Chan Lee,
Eunmi Chae,
Wieland Schöllkopf,
Bum Suk Zhao
Abstract:
Thermal energy atom scattering at a surface with grazing incidence conditions is an innovative method for investigating dispersive atom-surface interactions with potential application in quantum sensing interferometry. The complete establishment of this technique would require a detailed peak analysis, which has yet to be achieved. We examined peak-width fluctuations in atomic and molecular beams…
▽ More
Thermal energy atom scattering at a surface with grazing incidence conditions is an innovative method for investigating dispersive atom-surface interactions with potential application in quantum sensing interferometry. The complete establishment of this technique would require a detailed peak analysis, which has yet to be achieved. We examined peak-width fluctuations in atomic and molecular beams diffracted by a grating under grazing incidence conditions. Careful measurements and analyses of the diffraction patterns of He atoms and D$_2$ molecules from three square-wave gratings with different periods and radii of curvature enabled the identification of factors influencing the variations in the width as a function of the grazing incidence angle. The effects of macroscopic surface curvature, grating magnification, and beam emergence are substantial under these conditions but negligible for incidence angles close to the normal. Our results shed light on the phenomena occurring in grazing incidence thermal energy atom scattering.
△ Less
Submitted 17 January, 2024; v1 submitted 21 November, 2023;
originally announced November 2023.
-
Learning to Repair Software Vulnerabilities with Generative Adversarial Networks
Authors:
Jacob Harer,
Onur Ozdemir,
Tomo Lazovich,
Christopher P. Reale,
Rebecca L. Russell,
Louis Y. Kim,
Peter Chin
Abstract:
Motivated by the problem of automated repair of software vulnerabilities, we propose an adversarial learning approach that maps from one discrete source domain to another target domain without requiring paired labeled examples or source and target domains to be bijections. We demonstrate that the proposed adversarial learning approach is an effective technique for repairing software vulnerabilitie…
▽ More
Motivated by the problem of automated repair of software vulnerabilities, we propose an adversarial learning approach that maps from one discrete source domain to another target domain without requiring paired labeled examples or source and target domains to be bijections. We demonstrate that the proposed adversarial learning approach is an effective technique for repairing software vulnerabilities, performing close to seq2seq approaches that require labeled pairs. The proposed Generative Adversarial Network approach is application-agnostic in that it can be applied to other problems similar to code repair, such as grammar correction or sentiment translation.
△ Less
Submitted 28 October, 2018; v1 submitted 18 May, 2018;
originally announced May 2018.
-
Automated software vulnerability detection with machine learning
Authors:
Jacob A. Harer,
Louis Y. Kim,
Rebecca L. Russell,
Onur Ozdemir,
Leonard R. Kosta,
Akshay Rangamani,
Lei H. Hamilton,
Gabriel I. Centeno,
Jonathan R. Key,
Paul M. Ellingwood,
Erik Antelman,
Alan Mackay,
Marc W. McConley,
Jeffrey M. Opper,
Peter Chin,
Tomo Lazovich
Abstract:
Thousands of security vulnerabilities are discovered in production software each year, either reported publicly to the Common Vulnerabilities and Exposures database or discovered internally in proprietary code. Vulnerabilities often manifest themselves in subtle ways that are not obvious to code reviewers or the developers themselves. With the wealth of open source code available for analysis, the…
▽ More
Thousands of security vulnerabilities are discovered in production software each year, either reported publicly to the Common Vulnerabilities and Exposures database or discovered internally in proprietary code. Vulnerabilities often manifest themselves in subtle ways that are not obvious to code reviewers or the developers themselves. With the wealth of open source code available for analysis, there is an opportunity to learn the patterns of bugs that can lead to security vulnerabilities directly from data. In this paper, we present a data-driven approach to vulnerability detection using machine learning, specifically applied to C and C++ programs. We first compile a large dataset of hundreds of thousands of open-source functions labeled with the outputs of a static analyzer. We then compare methods applied directly to source code with methods applied to artifacts extracted from the build process, finding that source-based models perform better. We also compare the application of deep neural network models with more traditional models such as random forests and find the best performance comes from combining features learned by deep models with tree-based models. Ultimately, our highest performing model achieves an area under the precision-recall curve of 0.49 and an area under the ROC curve of 0.87.
△ Less
Submitted 2 August, 2018; v1 submitted 14 February, 2018;
originally announced March 2018.
-
Effect of rotational-state-dependent molecular alignment on the optical dipole force
Authors:
Lee Yeong Kim,
Ju Hyeon Lee,
Hye Ah Kim,
Sang Kyu Kwak,
Bretislav Friedrich,
Bum Suk Zhao
Abstract:
The properties of molecule-optical elements such as lenses or prisms based on the interaction of molecules with optical fields depend in a crucial way on the molecular quantum state and its alignment created by the optical field. However, in previous experimental studies, the effects of state-dependent alignment have never been included in estimates of the optical dipole force acting on the molecu…
▽ More
The properties of molecule-optical elements such as lenses or prisms based on the interaction of molecules with optical fields depend in a crucial way on the molecular quantum state and its alignment created by the optical field. However, in previous experimental studies, the effects of state-dependent alignment have never been included in estimates of the optical dipole force acting on the molecules while previous theoretical investigations took the state-dependent molecular alignment into account only implicitly. Herein, we consider the effects of molecular alignment explicitly and, to this end, introduce an effective polarizability which takes proper account of molecular alignment and is directly related to the alignment-dependent optical dipole force. We illustrate the significance of including molecular alignment in the optical dipole force by a trajectory study that compares previously used approximations with the present approach. The trajectory simulations were carried out for an ensemble of linear molecules subject to either propagating or standing-wave optical fields for a range of temperatures and laser intensities. The results demonstrate that the alignment-dependent effective polarizability can serve to provide correct estimates of the optical dipole force, on which a state-selection method applicable to nonpolar molecules could be based. We note that an analogous analysis of the forces acting on polar molecules subject to an inhomogeneous static electric field reveals a similarly strong dependence on molecular orientation.
△ Less
Submitted 1 June, 2016;
originally announced June 2016.
