Showing 1–50 of 58 results for author: Bishop, M

The integral Chow ring of $\overline{\mathcal{M}}_{1,3}$

Authors: Martin Bishop

Abstract: In this paper we compute the integral Chow ring of the moduli space of stable elliptic curves with three marked points by combining several patching techniques, including higher Chow groups with $\ell$-adic coefficients. In this paper we compute the integral Chow ring of the moduli space of stable elliptic curves with three marked points by combining several patching techniques, including higher Chow groups with $\ell$-adic coefficients. △ Less

Submitted 7 February, 2024; originally announced February 2024.

Comments: 18 pages, 2 figures

Few-Shot Detection of Machine-Generated Text using Style Representations

Authors: Rafael Rivera Soto, Kailin Koch, Aleem Khan, Barry Chen, Marcus Bishop, Nicholas Andrews

Abstract: The advent of instruction-tuned language models that convincingly mimic human writing poses a significant risk of abuse. However, such abuse may be counteracted with the ability to detect whether a piece of text was composed by a language model rather than a human author. Some previous approaches to this problem have relied on supervised methods by training on corpora of confirmed human- and machi… ▽ More The advent of instruction-tuned language models that convincingly mimic human writing poses a significant risk of abuse. However, such abuse may be counteracted with the ability to detect whether a piece of text was composed by a language model rather than a human author. Some previous approaches to this problem have relied on supervised methods by training on corpora of confirmed human- and machine- written documents. Unfortunately, model under-specification poses an unavoidable challenge for neural network-based detectors, making them brittle in the face of data shifts, such as the release of newer language models producing still more fluent text than the models used to train the detectors. Other approaches require access to the models that may have generated a document in question, which is often impractical. In light of these challenges, we pursue a fundamentally different approach not relying on samples from language models of concern at training time. Instead, we propose to leverage representations of writing style estimated from human-authored text. Indeed, we find that features effective at distinguishing among human authors are also effective at distinguishing human from machine authors, including state-of-the-art large language models like Llama-2, ChatGPT, and GPT-4. Furthermore, given a handful of examples composed by each of several specific language models of interest, our approach affords the ability to predict which model generated a given document. The code and data to reproduce our experiments are available at https://github.com/LLNL/LUAR/tree/main/fewshot_iclr2024. △ Less

Submitted 8 May, 2024; v1 submitted 12 January, 2024; originally announced January 2024.

The integral Chow ring of $\mathcal{M}_{1,n}$ for $n=3,\dots,10$

Authors: Martin Bishop

Abstract: We compute the integral Chow ring of the moduli stack of smooth elliptic curves with $n$ marked points for $3\leq n\leq 10$. We compute the integral Chow ring of the moduli stack of smooth elliptic curves with $n$ marked points for $3\leq n\leq 10$. △ Less

Submitted 19 November, 2023; originally announced November 2023.

Assessing the Security of GitHub Copilot Generated Code -- A Targeted Replication Study

Authors: Vahid Majdinasab, Michael Joshua Bishop, Shawn Rasheed, Arghavan Moradidakhel, Amjed Tahir, Foutse Khomh

Abstract: AI-powered code generation models have been develo** rapidly, allowing developers to expedite code generation and thus improve their productivity. These models are trained on large corpora of code (primarily sourced from public repositories), which may contain bugs and vulnerabilities. Several concerns have been raised about the security of the code generated by these models. Recent studies have… ▽ More AI-powered code generation models have been develo** rapidly, allowing developers to expedite code generation and thus improve their productivity. These models are trained on large corpora of code (primarily sourced from public repositories), which may contain bugs and vulnerabilities. Several concerns have been raised about the security of the code generated by these models. Recent studies have investigated security issues in AI-powered code generation tools such as GitHub Copilot and Amazon CodeWhisperer, revealing several security weaknesses in the code generated by these tools. As these tools evolve, it is expected that they will improve their security protocols to prevent the suggestion of insecure code to developers. This paper replicates the study of Pearce et al., which investigated security weaknesses in Copilot and uncovered several weaknesses in the code suggested by Copilot across diverse scenarios and languages (Python, C and Verilog). Our replication examines Copilot security weaknesses using newer versions of Copilot and CodeQL (the security analysis framework). The replication focused on the presence of security vulnerabilities in Python code. Our results indicate that, even with the improvements in newer versions of Copilot, the percentage of vulnerable code suggestions has reduced from 36.54% to 27.25%. Nonetheless, it remains evident that the model still suggests insecure code. △ Less

Submitted 18 November, 2023; originally announced November 2023.

Power spectrum estimation methods on intracluster medium surface brightness fluctuations

Authors: Mark Bishop, Yvette Perrott, Tulasi Parashar, Sean Oughton

Abstract: Accurate estimation of galaxy cluster masses is a central problem in cosmology. Turbulence is believed to introduce significant deviations from the hydrostatic mass estimates. Estimation of turbulence properties is complicated by projection of the 3D cluster onto the 2D plane of the sky, and is commonly done in the form of indirect probes from fluctuations in the X-ray surface brightness and Sunya… ▽ More Accurate estimation of galaxy cluster masses is a central problem in cosmology. Turbulence is believed to introduce significant deviations from the hydrostatic mass estimates. Estimation of turbulence properties is complicated by projection of the 3D cluster onto the 2D plane of the sky, and is commonly done in the form of indirect probes from fluctuations in the X-ray surface brightness and Sunyaev-Zeldovich effect maps. In this paper, we address this problem using simulations. We examine different methods for estimating the power spectrum on 2D projected fluctuation data, emulating data projected onto a 2D plane of the sky, and comparing them to the original, expected 3D power spectrum. Noise can contaminate the power spectrum of ICM observations, so we also briefly compare a few methods of reducing noise in the images for better spectral estimation. △ Less

Submitted 2 April, 2024; v1 submitted 10 October, 2023; originally announced October 2023.

Comments: 6 pages, 2 figures

Journal ref: Published in Proc. of the mm Universe 2023 conference, Grenoble (France), June 2023, published by F. Mayet et al. (Eds), EPJ Web of conferences, EDP Sciences, Volume 293 (2024)

Quantum gravity, the cosmological constant, and parity transformation

Authors: Michael Bishop, Peter Martin, Douglas Singleton

Abstract: One of the leading issues in quantum field theory and cosmology is the mismatch between the observed and calculated values for the cosmological constant in Einstein's field equations of up to 120 orders of magnitude. In this paper, we discuss new methods to potentially bridge this chasm using the generalized uncertainty principle (GUP). We find that if quantum gravity GUP models are the solution t… ▽ More One of the leading issues in quantum field theory and cosmology is the mismatch between the observed and calculated values for the cosmological constant in Einstein's field equations of up to 120 orders of magnitude. In this paper, we discuss new methods to potentially bridge this chasm using the generalized uncertainty principle (GUP). We find that if quantum gravity GUP models are the solution to this puzzle, then it may require the gravitationally modified position operator undergo a parity transformation at high energies. △ Less

Submitted 19 September, 2023; originally announced September 2023.

Comments: 10 pages, revtex-4, 0 figures, published in PLB

Journal ref: Phys.Lett.B 845 (2023) 138173

Narrow linewidth semiconductor lasers based on nonlinear self-injection locking

Authors: Andrew M. Bishop, Alexander L. Gaeta

Abstract: Self-injection locking techniques for stabilizing lasers have been developed using passive cavities to increase the effective lifetime of the laser cavity, thereby reducing the linewidth of the laser. We propose and demonstrate a new technique based on nonlinear self-injection locking (N-SIL) which we implement via feedback from the gain-narrowed Stokes mode of a fiber Brillouin oscillator. By blu… ▽ More Self-injection locking techniques for stabilizing lasers have been developed using passive cavities to increase the effective lifetime of the laser cavity, thereby reducing the linewidth of the laser. We propose and demonstrate a new technique based on nonlinear self-injection locking (N-SIL) which we implement via feedback from the gain-narrowed Stokes mode of a fiber Brillouin oscillator. By blue-shifting the Stokes field back to its pump frequency with an electro-optic modulator we realize recursive linewidth reduction that eliminates the phase drift caused by spontaneous emission noise. The fundamental linewidth limit is set by the spontaneous emission limit of the nonlinear oscillator, far lower than the spontaneous emission limit of a semiconductor laser. We demonstrate the power of this approach by achieving sub-hertz fundamental linewidth from the output of a commercial DFB laser and noise performance that significantly exceeds that of conventional SIL. We also and propose alternative fully-integrated designs in CMOS-compatible photonic platforms that allow for highly compact and robust implementations. △ Less

Submitted 18 September, 2023; originally announced September 2023.

Can Authorship Representation Learning Capture Stylistic Features?

Authors: Andrew Wang, Cristina Aggazzotti, Rebecca Kotula, Rafael Rivera Soto, Marcus Bishop, Nicholas Andrews

Abstract: Automatically disentangling an author's style from the content of their writing is a longstanding and possibly insurmountable problem in computational linguistics. At the same time, the availability of large text corpora furnished with author labels has recently enabled learning authorship representations in a purely data-driven manner for authorship attribution, a task that ostensibly depends to… ▽ More Automatically disentangling an author's style from the content of their writing is a longstanding and possibly insurmountable problem in computational linguistics. At the same time, the availability of large text corpora furnished with author labels has recently enabled learning authorship representations in a purely data-driven manner for authorship attribution, a task that ostensibly depends to a greater extent on encoding writing style than encoding content. However, success on this surrogate task does not ensure that such representations capture writing style since authorship could also be correlated with other latent variables, such as topic. In an effort to better understand the nature of the information these representations convey, and specifically to validate the hypothesis that they chiefly encode writing style, we systematically probe these representations through a series of targeted experiments. The results of these experiments suggest that representations learned for the surrogate authorship prediction task are indeed sensitive to writing style. As a consequence, authorship representations may be expected to be robust to certain kinds of data shift, such as topic drift over time. Additionally, our findings may open the door to downstream applications that require stylistic representations, such as style transfer. △ Less

Submitted 24 August, 2023; v1 submitted 22 August, 2023; originally announced August 2023.

Comments: appearing at TACL 2023

Minimal Lengths in 3D via the Generalized Uncertainty Principle

Authors: Michael Bishop, Joey Contreras, Peter Martin, Piero Nicolini, Douglas Singleton

Abstract: We investigate an extension of the Generalized Uncertainty Principle (GUP) in three dimensions by modifying the three dimensional position and momentum operators in a manner that remains coordinate-independent and retains as much of the standard position-momentum commutators as possible. Moreover, we bound the physical momentum which leads to an effective minimal length in every coordinate directi… ▽ More We investigate an extension of the Generalized Uncertainty Principle (GUP) in three dimensions by modifying the three dimensional position and momentum operators in a manner that remains coordinate-independent and retains as much of the standard position-momentum commutators as possible. Moreover, we bound the physical momentum which leads to an effective minimal length in every coordinate direction. The physical consequences of these modified operators are explored in two scenarios: (i) when a spherically-symmetric wave function is `compressed' into the smallest possible volume; (ii) when the momentum is directed in a single direction. In case (ii), we find that the three dimensional GUP exhibits interesting phenomena that do not occur in one dimension: the minimal distance in the direction parallel to a particle's momentum is different from the minimal distance in the orthogonal directions. △ Less

Submitted 7 July, 2023; originally announced July 2023.

Comments: 16 pages, no figures revtex4

Comments on the cosmological constant in generalized uncertainty models

Authors: Michael Bishop, Joey Contreras, Peter Martin, Douglas Singleton

Abstract: The existence of a small, non-zero cosmological constant is one of the major puzzles in fundamental physics. Naively, quantum field theory arguments would imply a cosmological constant which is up to 10$^{120}$ times larger than the observed one. It is believed a comprehensive theory of quantum gravity would resolve this enormous mismatch between theory and observation. In this work, we study the… ▽ More The existence of a small, non-zero cosmological constant is one of the major puzzles in fundamental physics. Naively, quantum field theory arguments would imply a cosmological constant which is up to 10$^{120}$ times larger than the observed one. It is believed a comprehensive theory of quantum gravity would resolve this enormous mismatch between theory and observation. In this work, we study the ability of generalized uncertainty principle (GUP) models, which are phenomenologically motivated models of quantum gravity, to address the cosmological constant problem. In particular, we focus on how these GUP models may change the phase space of QFT, and how this affects the momentum space integration of the zero-point energies of normal modes of fields. We point out several issues that make it unlikely that GUP models, in their current form, would be able to adequately address the cosmological constant problem. △ Less

Submitted 23 August, 2022; originally announced September 2022.

Comments: 10 pages, revtex, no figures. Published in Frontiers in Astronomy and Space Science. This article is part of the Research Topic "Generalized Uncertainty Relations: Existing Paradigms and New Approaches" edited by Dr. Matthew Lake

Journal ref: Front. Astron. Space Sci. 9, 23 (2022)

The more things change the more they stay the same: Minimum lengths with unmodified uncertainty principle and dispersion relation

Authors: Michael Bishop, Joey Contreras, Douglas Singleton

Abstract: Broad arguments indicate that quantum gravity should have a minimal length scale. In this essay we construct a minimum length model by generalizing the time-position and energy-momentum operators while kee** much of the structure of quantum mechanics and relativity intact: the standard position-momentum commutator, the special relativistic time-position, and energy-momentum relationships all rem… ▽ More Broad arguments indicate that quantum gravity should have a minimal length scale. In this essay we construct a minimum length model by generalizing the time-position and energy-momentum operators while kee** much of the structure of quantum mechanics and relativity intact: the standard position-momentum commutator, the special relativistic time-position, and energy-momentum relationships all remain the same. Since the time-position and energy-momentum relationships for the modified operators remains the same, we retain a form of Lorentz symmetry. This avoids the constraints on these theories coming from lack of photon dispersion while holding the potential to address the Greisen-Zatsepin-Kuzmin (GZK) puzzle of ultra high energy cosmic rays. △ Less

Submitted 11 May, 2022; originally announced May 2022.

Comments: 10 pages, 2 figures, revtex4. 2nd place award for the 2022 Gravity Research Foundation essay competition. To be published IJMPD

GlacierNet2: A Hybrid Multi-Model Learning Architecture for Alpine Glacier Map**

Authors: Zhiyuan Xie, Umesh K. Haritashya, Vijayan K. Asari, Michael P. Bishop, Jeffrey S. Kargel, Theus H. Aspiras

Abstract: In recent decades, climate change has significantly affected glacier dynamics, resulting in mass loss and an increased risk of glacier-related hazards including supraglacial and proglacial lake development, as well as catastrophic outburst flooding. Rapidly changing conditions dictate the need for continuous and detailed observations and analysis of climate-glacier dynamics. Thematic and quantitat… ▽ More In recent decades, climate change has significantly affected glacier dynamics, resulting in mass loss and an increased risk of glacier-related hazards including supraglacial and proglacial lake development, as well as catastrophic outburst flooding. Rapidly changing conditions dictate the need for continuous and detailed observations and analysis of climate-glacier dynamics. Thematic and quantitative information regarding glacier geometry is fundamental for understanding climate forcing and the sensitivity of glaciers to climate change, however, accurately map** debris-cover glaciers (DCGs) is notoriously difficult based upon the use of spectral information and conventional machine-learning techniques. The objective of this research is to improve upon an earlier proposed deep-learning-based approach, GlacierNet, which was developed to exploit a convolutional neural-network segmentation model to accurately outline regional DCG ablation zones. Specifically, we developed an enhanced GlacierNet2 architecture thatincorporates multiple models, automatic post-processing, and basin-level hydrological flow techniques to improve the map** of DCGs such that it includes both the ablation and accumulation zones. Experimental evaluations demonstrate that GlacierNet2 improves the estimation of the ablation zone and allows a high level of intersection over union (IOU: 0.8839) score. The proposed architecture provides complete glacier (both accumulation and ablation zone) outlines at regional scales, with an overall IOU score of 0.8619. This is a crucial first step in automating complete glacier map** that can be used for accurate glacier modeling or mass-balance analysis. △ Less

Submitted 29 July, 2022; v1 submitted 6 April, 2022; originally announced April 2022.

Journal ref: International Journal of Applied Earth Observation and Geoinformation, 2022

A Subtle Aspect of Minimal Lengths in the Generalized Uncertainty Principle

Authors: Michael Bishop, Joey Contreras, Douglas Singleton

Abstract: In this work, we point out an overlooked and subtle feature of the generalized uncertainty principle (GUP) approach to quantizing gravity: namely that different pairs of modified operators with the same modified commutator, $[\hat{X},\hat{P}] = i \hbar (1+βp^2)$, may have different physical consequences such as having no minimal length at all. These differences depend on how the position and/or mo… ▽ More In this work, we point out an overlooked and subtle feature of the generalized uncertainty principle (GUP) approach to quantizing gravity: namely that different pairs of modified operators with the same modified commutator, $[\hat{X},\hat{P}] = i \hbar (1+βp^2)$, may have different physical consequences such as having no minimal length at all. These differences depend on how the position and/or momentum operators are modified rather than only on the resulting modified commutator. This provides guidance when constructing GUP models since it distinguishes those GUPs that have a minimal length scale, as suggested by some broad arguments about quantum gravity, versus GUPs without a minimal length scale. △ Less

Submitted 20 March, 2022; originally announced March 2022.

Comments: 14 pages, 3 figures. Published in Universe

Journal ref: Universe 2022, 8(3), 192

Explaining RADAR features for detecting spoofing attacks in Connected Autonomous Vehicles

Authors: Nidhi Rastogi, Sara Rampazzi, Michael Clifford, Miriam Heller, Matthew Bishop, Karl Levitt

Abstract: Connected autonomous vehicles (CAVs) are anticipated to have built-in AI systems for defending against cyberattacks. Machine learning (ML) models form the basis of many such AI systems. These models are notorious for acting like black boxes, transforming inputs into solutions with great accuracy, but no explanations support their decisions. Explanations are needed to communicate model performance,… ▽ More Connected autonomous vehicles (CAVs) are anticipated to have built-in AI systems for defending against cyberattacks. Machine learning (ML) models form the basis of many such AI systems. These models are notorious for acting like black boxes, transforming inputs into solutions with great accuracy, but no explanations support their decisions. Explanations are needed to communicate model performance, make decisions transparent, and establish trust in the models with stakeholders. Explanations can also indicate when humans must take control, for instance, when the ML model makes low confidence decisions or offers multiple or ambiguous alternatives. Explanations also provide evidence for post-incident forensic analysis. Research on explainable ML to security problems is limited, and more so concerning CAVs. This paper surfaces a critical yet under-researched sensor data \textit{uncertainty} problem for training ML attack detection models, especially in highly mobile and risk-averse platforms such as autonomous vehicles. We present a model that explains \textit{certainty} and \textit{uncertainty} in sensor input -- a missing characteristic in data collection. We hypothesize that model explanation is inaccurate for a given system without explainable input data quality. We estimate \textit{uncertainty} and mass functions for features in radar sensor data and incorporate them into the training model through experimental evaluation. The mass function allows the classifier to categorize all spoofed inputs accurately with an incorrect class label. △ Less

Submitted 28 February, 2022; originally announced March 2022.

Comments: Accepted at the AAAI 2022 Workshop on Explainable Agency in Artificial Intelligence Workshop, Virtual. 8 pages, 3 Figures, 4 tables

MSC Class: 68M25; 60B11; 68T05

An automated near-real time computational method for induction and treatment of scar-related ventricular tachycardias

Authors: Fernando O. Campos, Aurel Neic, Caroline Mendonca Costa, John Whitaker, Mark O'Neill, Reza Razavi, Christopher A. Rinaldi, Daniel Scherr, Steven A. Niederer, Gernot Plank, Martin J. Bishop

Abstract: Catheter ablation is currently the only curative treatment for scar-related ventricular tachycardias (VTs). However, not only are ablation procedures long, with relatively high risk, but success rates are punitively low, with frequent VT recurrence. Personalized in-silico approaches have the opportunity to address these limitations. However, state-of-the-art reaction diffusion (R-D) simulations of… ▽ More Catheter ablation is currently the only curative treatment for scar-related ventricular tachycardias (VTs). However, not only are ablation procedures long, with relatively high risk, but success rates are punitively low, with frequent VT recurrence. Personalized in-silico approaches have the opportunity to address these limitations. However, state-of-the-art reaction diffusion (R-D) simulations of VT induction and subsequent circuits used for in-silico ablation target identification require long execution times, along with vast computational resources, which are incompatible with the clinical workflow. Here, we present the Virtual Induction and Treatment of Arrhythmias (VITA), a novel, rapid and fully automated computational approach that uses reaction-Eikonal methodology to induce VT and identify subsequent ablation targets. The rationale for VITA is based on finding isosurfaces associated with an activation wavefront that splits in the ventricles due to the presence of an isolated isthmus of conduction within the scar; once identified, each isthmus may be assessed for their vulnerability to sustain a reentrant circuit, and the corresponding exit site automatically identified for potential ablation targeting. VITA was tested on a virtual cohort of 7 post-infarcted porcine hearts and the results compared to R-D simulations. Using only a standard desktop machine, VITA could detect all scar-related VTs, simulating activation time maps and ECGs (for clinical comparison) as well as computing ablation targets in 48 minutes. The comparable VTs probed by the R-D simulations took 68.5 hours on 256 cores of high-performance computing infrastructure. The set of lesions computed by VITA was shown to render the ventricular model VT-free. VITA could be used in near real-time as a complementary modality aiding in clinical decision-making in the treatment of post-infarction VTs. △ Less

Submitted 27 August, 2021; originally announced August 2021.

Monte Carlo computer investigations of higher generation ideal dendrimers

Authors: M. Jura, M. Bishop, B. Thrope, R. de Regt

Abstract: The properties of ideal tri-functional dendrimers with forty-five, ninety-three and one hundred and eighty-nine branches are investigated. Three methods are employed to calculate the mean-square radius of gyration, $g$-ratios, asphericity, shape parameters and form factor. These methods include a Kirchhoff matrix eigenvalue technique, the graph theory approach of Benhamou et al. (2004), and Monte… ▽ More The properties of ideal tri-functional dendrimers with forty-five, ninety-three and one hundred and eighty-nine branches are investigated. Three methods are employed to calculate the mean-square radius of gyration, $g$-ratios, asphericity, shape parameters and form factor. These methods include a Kirchhoff matrix eigenvalue technique, the graph theory approach of Benhamou et al. (2004), and Monte Carlo simulations using a growth algorithm. A novel technique for counting paths in the graph representation of the dendrimers is presented. All the methods are in excellent agreement with each other and with available theoretical predictions. Dendrimers become more symmetrical as the generation and the number of branches increase. △ Less

Submitted 24 June, 2021; originally announced June 2021.

Comments: 13 pages, 4 figures, 8 tables

Journal ref: Condens. Matter Phys., 2021, vol. 24, No. 2, 23301

A Deep Metric Learning Approach to Account Linking

Authors: Aleem Khan, Elizabeth Fleming, Noah Schofield, Marcus Bishop, Nicholas Andrews

Abstract: We consider the task of linking social media accounts that belong to the same author in an automated fashion on the basis of the content and metadata of their corresponding document streams. We focus on learning an embedding that maps variable-sized samples of user activity -- ranging from single posts to entire months of activity -- to a vector space, where samples by the same author map to nearb… ▽ More We consider the task of linking social media accounts that belong to the same author in an automated fashion on the basis of the content and metadata of their corresponding document streams. We focus on learning an embedding that maps variable-sized samples of user activity -- ranging from single posts to entire months of activity -- to a vector space, where samples by the same author map to nearby points. The approach does not require human-annotated data for training purposes, which allows us to leverage large amounts of social media content. The proposed model outperforms several competitive baselines under a novel evaluation framework modeled after established recognition benchmarks in other domains. Our method achieves high linking accuracy, even with small samples from accounts not seen at training time, a prerequisite for practical applications of the proposed linking framework. △ Less

Submitted 15 May, 2021; originally announced May 2021.

Comments: 13 pages; to be published in NAACL 2021

Quincke oscillations of colloids at planar electrodes

Authors: Zhengyan Zhang, Hang Yuan, Yong Dou, Monica Olvera de la Cruz, Kyle J. M. Bishop

Abstract: Dielectric particles in weakly conducting fluids rotate spontaneously when subject to strong electric fields. Such Quincke rotation near a plane electrode leads to particle translation that enables physical models of active matter. Here, we show that Quincke rollers can also exhibit oscillatory dynamics, whereby particles move back and forth about a fixed location. We explain how oscillations aris… ▽ More Dielectric particles in weakly conducting fluids rotate spontaneously when subject to strong electric fields. Such Quincke rotation near a plane electrode leads to particle translation that enables physical models of active matter. Here, we show that Quincke rollers can also exhibit oscillatory dynamics, whereby particles move back and forth about a fixed location. We explain how oscillations arise for micron-scale particles commensurate with the thickness of a field-induced boundary layer in the nonpolar electrolyte. This work enables the design of colloidal oscillators. △ Less

Submitted 9 February, 2021; originally announced February 2021.

Journal ref: Phys. Rev. Lett. 126, 258001 (2021)

Patterns that persist: Heritable information in stochastic dynamics

Authors: Peter M. Tzelios, Kyle J. M. Bishop

Abstract: Life on earth is distinguished by long-lived correlations in time. The patterns of material organization that characterize living organisms today are contingent on events that occurred billions of years ago. This contingency is a necessary component of Darwinian evolution: patterns in the present inherit some of their features from those in the past. Despite its central role in biology, heritable… ▽ More Life on earth is distinguished by long-lived correlations in time. The patterns of material organization that characterize living organisms today are contingent on events that occurred billions of years ago. This contingency is a necessary component of Darwinian evolution: patterns in the present inherit some of their features from those in the past. Despite its central role in biology, heritable information is difficult to recognize in prebiotic systems described in the language of chemistry or physics. Here, we consider one such description based on continuous-time Markov processes and investigate the persistence of heritable information within large sets of dynamical systems. While the microscopic state of each system fluctuates incessantly, there exist few systems that relax slowly to their stationary distribution over much longer times. These systems, selected for their persistence, are further distinguished by their mesoscopic organization, which allows for accurate course grained descriptions of their dynamics at long times. The slow relaxation of these stable patterns is made possible by dissipative currents fueled by thermodynamic gradients present in the surrounding reservoirs. We show how the rate of entropy production within a system sets an upper bound on the lifetime of its persistent patterns. We also consider the probability of finding persistence within large sets of dynamical systems. We show that physical constraints based on continuity and locality can strongly influence the probability of persistence and its dependence on system size. Finally, we describe how heritable information can be quantified in practice using universal compression algorithms. We demonstrate this approach on an experimental system of magnetically-driven, colloidal rollers and discuss the application of these methods to origins of life research. △ Less

Submitted 5 January, 2021; originally announced January 2021.

Comments: 30 pages, 11 Figures

Reconciling a quantum gravity minimal length with lack of photon dispersion

Authors: Michael Bishop, Joey Contreras, Jaeyeong Lee, Douglas Singleton

Abstract: Generic arguments lead to the idea that quantum gravity has a minimal length scale. A possible observational signal of such a minimal length scale is that photons should exhibit dispersion. In 2009 the observation of a short gamma ray burst seemed to bound the minimal length scale to distances smaller than the Planck length, implying that spacetime appeared continuous to distances below the Planck… ▽ More Generic arguments lead to the idea that quantum gravity has a minimal length scale. A possible observational signal of such a minimal length scale is that photons should exhibit dispersion. In 2009 the observation of a short gamma ray burst seemed to bound the minimal length scale to distances smaller than the Planck length, implying that spacetime appeared continuous to distances below the Planck length. This poses a challenge for such minimal distance models. Here we propose a modification of the position and momentum operators, ${\hat x}$ and ${\hat p}$, which lead to a minimal length scale, but preserve the photon energy-momentum relationship $E = p c$. In this way there is no dispersion of photons with different energies. This can be accomplished without modifying the commutation relationship $[{\hat x}, {\hat p}] = i \hbar$. △ Less

Submitted 29 March, 2021; v1 submitted 25 September, 2020; originally announced September 2020.

Comments: 17 pages and 4 figures. Added references and discussion to address reviewer comments. To be published in PLB

Artificial Intelligence is stupid and causal reasoning won't fix it

Authors: John Mark Bishop

Abstract: Artificial Neural Networks have reached Grandmaster and even super-human performance across a variety of games: from those involving perfect-information (such as Go) to those involving imperfect-information (such as Starcraft). Such technological developments from AI-labs have ushered concomitant applications across the world of business - where an AI brand tag is fast becoming ubiquitous. A corol… ▽ More Artificial Neural Networks have reached Grandmaster and even super-human performance across a variety of games: from those involving perfect-information (such as Go) to those involving imperfect-information (such as Starcraft). Such technological developments from AI-labs have ushered concomitant applications across the world of business - where an AI brand tag is fast becoming ubiquitous. A corollary of such widespread commercial deployment is that when AI gets things wrong - an autonomous vehicle crashes; a chatbot exhibits racist behaviour; automated credit scoring processes discriminate on gender etc. - there are often significant financial, legal and brand consequences and the incident becomes major news. As Judea Pearl sees it, the underlying reason for such mistakes is that, 'all the impressive achievements of deep learning amount to just curve fitting'. The key, Judea Pearl suggests, is to replace reasoning by association with causal-reasoning - the ability to infer causes from observed phenomena. It is a point that was echoed by Gary Marcus and Ernest Davis in a recent piece for the New York Times: 'we need to stop building computer systems that merely get better and better at detecting statistical patterns in data sets - often using an approach known as Deep Learning - and start building computer systems that from the moment of their assembly innately grasp three basic concepts: time, space and causality'. In this paper, foregrounding what in 1949 Gilbert Ryle termed a category mistake, I will offer an alternative explanation for AI errors: it is not so much that AI machinery cannot grasp causality, but that AI machinery - qua computation - cannot understand anything at all. △ Less

Submitted 20 July, 2020; originally announced August 2020.

Replication Markets: Results, Lessons, Challenges and Opportunities in AI Replication

Authors: Yang Liu, Michael Gordon, Juntao Wang, Michael Bishop, Yiling Chen, Thomas Pfeiffer, Charles Twardy, Domenico Viganola

Abstract: The last decade saw the emergence of systematic large-scale replication projects in the social and behavioral sciences, (Camerer et al., 2016, 2018; Ebersole et al., 2016; Klein et al., 2014, 2018; Collaboration, 2015). These projects were driven by theoretical and conceptual concerns about a high fraction of "false positives" in the scientific publications (Ioannidis, 2005) (and a high prevalence… ▽ More The last decade saw the emergence of systematic large-scale replication projects in the social and behavioral sciences, (Camerer et al., 2016, 2018; Ebersole et al., 2016; Klein et al., 2014, 2018; Collaboration, 2015). These projects were driven by theoretical and conceptual concerns about a high fraction of "false positives" in the scientific publications (Ioannidis, 2005) (and a high prevalence of "questionable research practices" (Simmons, Nelson, and Simonsohn, 2011). Concerns about the credibility of research findings are not unique to the behavioral and social sciences; within Computer Science, Artificial Intelligence (AI) and Machine Learning (ML) are areas of particular concern (Lucic et al., 2018; Freire, Bonnet, and Shasha, 2012; Gundersen and Kjensmo, 2018; Henderson et al., 2018). Given the pioneering role of the behavioral and social sciences in the promotion of novel methodologies to improve the credibility of research, it is a promising approach to analyze the lessons learned from this field and adjust strategies for Computer Science, AI and ML In this paper, we review approaches used in the behavioral and social sciences and in the DARPA SCORE project. We particularly focus on the role of human forecasting of replication outcomes, and how forecasting can leverage the information gained from relatively labor and resource-intensive replications. We will discuss opportunities and challenges of using these approaches to monitor and improve the credibility of research areas in Computer Science, AI, and ML. △ Less

Submitted 9 May, 2020; originally announced May 2020.

Comments: Appeared at AAAI workshop on Reproducible AI (RAI), 2020

Frequency-Domain Quantum Interference with Correlated Photons from an Integrated Microresonator

Authors: Chaitali Joshi, Alessandro Farsi, Avik Dutt, Bok Young Kim, Xingchen Ji, Yun Zhao, Andrew M. Bishop, Michal Lipson, Alexander L. Gaeta

Abstract: Frequency encoding of quantum information together with fiber and integrated photonic technologies can significantly reduce the complexity and resource requirements for realizing all-photonic quantum networks. The key challenge for such frequency domain processing of single photons is to realize coherent and selective interactions between quantum optical fields of different frequencies over a rang… ▽ More Frequency encoding of quantum information together with fiber and integrated photonic technologies can significantly reduce the complexity and resource requirements for realizing all-photonic quantum networks. The key challenge for such frequency domain processing of single photons is to realize coherent and selective interactions between quantum optical fields of different frequencies over a range of bandwidths. Here, we report frequency-domain Hong-Ou-Mandel interference with spectrally distinct photons generated from a chip-based microresonator. We use four-wave mixing to implement an active frequency beam-splitter and achieve interference visibilities of $0.95 \pm 0.02$. Our work establishes four-wave mixing as a tool for selective high-fidelity two-photon operations in the frequency domain which, combined with integrated single-photon sources, provides a building block for frequency-multiplexed photonic quantum networks. △ Less

Submitted 13 March, 2020; originally announced March 2020.

Journal ref: Phys. Rev. Lett. 124, 143601 (2020)

Swelling cholesteric liquid crystal shells to direct colloids at the interface

Authors: Lisa Tran, Kyle J. M. Bishop

Abstract: Cholesteric liquid crystals can exhibit spatial patterns in molecular alignment at interfaces that can be exploited for particle assembly. These patterns emerge from the competition between bulk and surface energies, tunable with the system geometry. In this work, we use the osmotic swelling of cholesteric double emulsions to assemble colloidal particles through a pathway-dependent process. Partic… ▽ More Cholesteric liquid crystals can exhibit spatial patterns in molecular alignment at interfaces that can be exploited for particle assembly. These patterns emerge from the competition between bulk and surface energies, tunable with the system geometry. In this work, we use the osmotic swelling of cholesteric double emulsions to assemble colloidal particles through a pathway-dependent process. Particles can be repositioned from a surface-mediated to an elasticity-mediated state through dynamically thinning the cholesteric shell at a rate comparable to that of colloidal adsorption. By tuning the balance between surface and bulk energies with the system geometry, colloidal assemblies on the cholesteric interface can be molded by the underlying elastic field to form linear aggregates. The transition of adsorbed particles from surface regions with homeotropic anchoring to defect regions is accompanied by a reduction in particle mobility. The arrested assemblies subsequently map out and stabilize topological defects. These results demonstrate the kinetic arrest of interfacial particles within definable patterns by regulating the energetic frustration within cholesterics. This work highlights the importance of kinetic pathways for particle assembly in liquid crystals, of relevance to optical and energy applications. △ Less

Submitted 4 January, 2020; originally announced January 2020.

Comments: 21 pages, 9 figures total, including 4 supplemental figures

Journal ref: ACS Nano 14 (5), 5459-5467 (2020)

Modified commutators are not sufficient to determine a quantum gravity minimal length scale

Authors: Michael Bishop, Jaeyeong Lee, Douglas Singleton

Abstract: In quantum gravity it is generally thought that a modified commutator of the form $[{\hat x}, {\hat p}] = i \hbar (1 + βp^2)$ is sufficient to give rise to a minimum length scale. We test this assumption and find that different pairs of modified operators can lead to the same modified commutator and yet give different or even no minimal length. The conclusion is that the modification of the operat… ▽ More In quantum gravity it is generally thought that a modified commutator of the form $[{\hat x}, {\hat p}] = i \hbar (1 + βp^2)$ is sufficient to give rise to a minimum length scale. We test this assumption and find that different pairs of modified operators can lead to the same modified commutator and yet give different or even no minimal length. The conclusion is that the modification of the operators is the main factor in determining whether there is a minimal length. This fact - that it is the specific form of the modified operators which determine the existence or not of a minimal length scale - can be used to keep or reject specific modifications of the position and momentum operators in theory of quantum gravity. △ Less

Submitted 9 January, 2020; v1 submitted 7 November, 2019; originally announced November 2019.

Comments: 14 pages and 3 figures. Revised to address referee comments. References added. To be published in PLB

Journal ref: Phys.Lett.B 802, 135209 (2020)

A new method for flow-based network intrusion detection using the inverse Potts model

Authors: Camila Pontes, Manuela Souza, João Gondim, Matt Bishop, Marcelo Marotta

Abstract: Network Intrusion Detection Systems (NIDS) play an important role as tools for identifying potential network threats. In the context of ever-increasing traffic volume on computer networks, flow-based NIDS arise as good solutions for real-time traffic classification. In recent years, different flow-based classifiers have been proposed using Machine Learning (ML) algorithms. Nevertheless, classical… ▽ More Network Intrusion Detection Systems (NIDS) play an important role as tools for identifying potential network threats. In the context of ever-increasing traffic volume on computer networks, flow-based NIDS arise as good solutions for real-time traffic classification. In recent years, different flow-based classifiers have been proposed using Machine Learning (ML) algorithms. Nevertheless, classical ML-based classifiers have some limitations. For instance, they require large amounts of labeled data for training, which might be difficult to obtain. Additionally, most ML-based classifiers are not capable of domain adaptation, i.e. after being trained on an specific data distribution, they are not general enough to be applied to other related data distributions. And, finally, many of the models inferred by these algorithms are black boxes, which do not provide explainable results. To overcome these limitations, we propose a new algorithm, called Energy-based Flow Classifier (EFC). This anomaly-based classifier uses inverse statistics to infer a statistical model based on labeled benign examples. We show that EFC is capable of accurately performing binary flow classification and is more adaptable to different data distributions than classical ML-based classifiers. Given the positive results obtained on three different datasets (CIDDS-001, CICIDS17 and CICDDoS19), we consider EFC to be a promising algorithm to perform robust flow-based traffic classification. △ Less

Submitted 23 September, 2021; v1 submitted 16 October, 2019; originally announced October 2019.

Journal ref: IEEE Transactions on Network and Service Management (Volume: 18, Issue: 2, June 2021, Page(s): 1125-1136) Date of Publication: 26 April 2021

Learning Invariant Representations of Social Media Users

Authors: Nicholas Andrews, Marcus Bishop

Abstract: The evolution of social media users' behavior over time complicates user-level comparison tasks such as verification, classification, clustering, and ranking. As a result, naïve approaches may fail to generalize to new users or even to future observations of previously known users. In this paper, we propose a novel procedure to learn a map** from short episodes of user activity on social media t… ▽ More The evolution of social media users' behavior over time complicates user-level comparison tasks such as verification, classification, clustering, and ranking. As a result, naïve approaches may fail to generalize to new users or even to future observations of previously known users. In this paper, we propose a novel procedure to learn a map** from short episodes of user activity on social media to a vector space in which the distance between points captures the similarity of the corresponding users' invariant features. We fit the model by optimizing a surrogate metric learning objective over a large corpus of unlabeled social media content. Once learned, the map** may be applied to users not seen at training time and enables efficient comparisons of users in the resulting vector space. We present a comprehensive evaluation to validate the benefits of the proposed approach using data from Reddit, Twitter, and Wikipedia. △ Less

Submitted 11 October, 2019; originally announced October 2019.

Comments: 12 pages, 3 figures; to be published in EMNLP 2019

Autonomous navigation of shape-shifting microswimmers

Authors: Yong Dou, Kyle J. M. Bishop

Abstract: We describe a method for programming the autonomous navigation of active colloidal particles in response to spatial gradients in a scalar stimulus. Functional behaviors such as positive or negative chemotaxis are encoded in the particle shape, which responds to the local stimulus and directs self-propelled particle motions. We demonstrate this approach using a physical model of stimuli-responsive… ▽ More We describe a method for programming the autonomous navigation of active colloidal particles in response to spatial gradients in a scalar stimulus. Functional behaviors such as positive or negative chemotaxis are encoded in the particle shape, which responds to the local stimulus and directs self-propelled particle motions. We demonstrate this approach using a physical model of stimuli-responsive clusters of self-phoretic spheres. We show how multiple autonomous behaviors can be achieved by designing the particle geometry and its stimulus response. △ Less

Submitted 15 August, 2019; originally announced August 2019.

Journal ref: Phys. Rev. Research 1, 032030 (2019)

Compact Optical Atomic Clock Based on a Two-Photon Transition in Rubidium

Authors: Kyle W. Martin, Gretchen Phelps, Nathan D. Lemke, Matthew S. Bigelow, Benjamin Stuhl, Michael Wojcik, Michael Holt, Ian Coddington, Michael W. Bishop, Johh H. Burke

Abstract: Extra-laboratory atomic clocks are necessary for a wide array of applications (e.g. satellite-based navigation and communication). Building upon existing vapor cell and laser technologies, we describe an optical atomic clock, designed around a simple and manufacturable architecture, that utilizes the 778~nm two-photon transition in rubidium and yields fractional frequency instabilities of… ▽ More Extra-laboratory atomic clocks are necessary for a wide array of applications (e.g. satellite-based navigation and communication). Building upon existing vapor cell and laser technologies, we describe an optical atomic clock, designed around a simple and manufacturable architecture, that utilizes the 778~nm two-photon transition in rubidium and yields fractional frequency instabilities of $3\times10^{-13}/\sqrt{τ(s)}$ for $τ$ from 1~s to 10000~s. We present a complete stability budget for this system and explore the required conditions under which a fractional frequency instability of $1\times 10^{-15}$ can be maintained on long timescales. We provide precise characterization of the leading sensitivities to external processes including magnetic fields and fluctuations of the vapor cell temperature and 778~nm laser power. The system is constructed primarily from commercially-available components, an attractive feature from the standpoint of commercialization and deployment of optical frequency standards. △ Less

Submitted 26 March, 2019; originally announced March 2019.

Comments: 9 pages, 6 figures

Journal ref: Phys. Rev. Applied 9, 014019, 2018

One-Loop Fluctuation Entropy of Charge Inversion in DNA

Authors: Matthew D. Sievert, Marilyn F. Bishop, Tom McMullen

Abstract: Experiments have revealed correlation-driven behavior of DNA in charged solutions, including charge inversion and condensation. This paper presents calculations of a lattice-gas model of charge inversion for the adsorption of charged dimers on DNA. Each adsorption site is assumed to have either a vacancy or a positively-charged dimer attached with the dimer oriented either parallel or perpendicula… ▽ More Experiments have revealed correlation-driven behavior of DNA in charged solutions, including charge inversion and condensation. This paper presents calculations of a lattice-gas model of charge inversion for the adsorption of charged dimers on DNA. Each adsorption site is assumed to have either a vacancy or a positively-charged dimer attached with the dimer oriented either parallel or perpendicular to the double helix DNA chain. The entropy and charge distributions of these three species are calculated including the lowest order fluctuation corrections to mean-field theory. We find that the inclusion of the fluctuation terms has a significant effect on the entropy, primarily in the regime where the dimers are repelled from the DNA molecule and compete with the chemical potential in solution. △ Less

Submitted 14 February, 2019; originally announced February 2019.

Comments: 30 pages, 26 figures

Learning retrosynthetic planning through self-play

Authors: John S. Schreck, Connor W. Coley, Kyle J. M. Bishop

Abstract: The problem of retrosynthetic planning can be framed as one player game, in which the chemist (or a computer program) works backwards from a molecular target to simpler starting materials though a series of choices regarding which reactions to perform. This game is challenging as the combinatorial space of possible choices is astronomical, and the value of each choice remains uncertain until the s… ▽ More The problem of retrosynthetic planning can be framed as one player game, in which the chemist (or a computer program) works backwards from a molecular target to simpler starting materials though a series of choices regarding which reactions to perform. This game is challenging as the combinatorial space of possible choices is astronomical, and the value of each choice remains uncertain until the synthesis plan is completed and its cost evaluated. Here, we address this problem using deep reinforcement learning to identify policies that make (near) optimal reaction choices during each step of retrosynthetic planning. Using simulated experience or self-play, we train neural networks to estimate the expected synthesis cost or value of any given molecule based on a representation of its molecular structure. We show that learned policies based on this value network outperform heuristic approaches in synthesizing unfamiliar molecules from available starting materials using the fewest number of reactions. We discuss how the learned policies described here can be incorporated into existing synthesis planning tools and how they can be adapted to changes in the synthesis cost objective or material availability. △ Less

Submitted 19 January, 2019; originally announced January 2019.

Modified commutation relationships from the Berry-Keating program

Authors: Michael Bishop, Erick Aiken, Douglas Singleton

Abstract: Current approaches to quantum gravity suggest there should be a modification of the standard quantum mechanical commutator, $[{\hat x} , {\hat p}] = i \hbar$. Typical modifications are phenomenological and designed to result in a minimal length scale. As a motivating principle for the modification of the position and momentum commutator, we assume the validity of a version of the Bender-Brody-Müll… ▽ More Current approaches to quantum gravity suggest there should be a modification of the standard quantum mechanical commutator, $[{\hat x} , {\hat p}] = i \hbar$. Typical modifications are phenomenological and designed to result in a minimal length scale. As a motivating principle for the modification of the position and momentum commutator, we assume the validity of a version of the Bender-Brody-Müller variant of the Berry-Keating approach to the Riemann hypothesis. We arrive at a family of modified position and momentum operators, and their associated modified commutator, which lead to a minimal length scale. Additionally, this larger family generalizes the Bender-Brody-Müller approach to the Riemann hypothesis. △ Less

Submitted 25 January, 2019; v1 submitted 30 September, 2018; originally announced October 2018.

Comments: 14 pages, 0 figures revtex4. Version published in PRD

Journal ref: Phys. Rev. D 99, 026012 (2019)

Intrinsically ultrastrong plasmon-exciton interactions in crystallized films of carbon nanotubes

Authors: Po-Hsun Ho, Damon B. Farmer, George S. Tulevski, Shu-Jen Han, Douglas M. Bishop, Lynne M. Gignac, Jim Bucchignano, Phaedon Avouris, Abram L. Falk

Abstract: In cavity quantum electrodynamics, optical emitters that are strongly coupled to cavities give rise to polaritons with characteristics of both the emitters and the cavity excitations. We show that carbon nanotubes can be crystallized into chip-scale, two-dimensionally ordered films and that this new material enables intrinsically ultrastrong emitter-cavity interactions: rather than interacting wit… ▽ More In cavity quantum electrodynamics, optical emitters that are strongly coupled to cavities give rise to polaritons with characteristics of both the emitters and the cavity excitations. We show that carbon nanotubes can be crystallized into chip-scale, two-dimensionally ordered films and that this new material enables intrinsically ultrastrong emitter-cavity interactions: rather than interacting with external cavities, nanotube excitons couple to the near-infrared plasmon resonances of the nanotubes themselves. Our polycrystalline nanotube films have a hexagonal crystal structure, ~25 nm domains, and a 1.74 nm lattice constant. With this extremely high nanotube density and nearly ideal plasmon-exciton spatial overlap, plasmon-exciton coupling strengths reach 0.5 eV, which is 75% of the bare exciton energy and a near record for room-temperature ultrastrong coupling. Crystallized nanotube films represent a milestone in nanomaterials assembly and provide a compelling foundation for high-ampacity conductors, low-power optical switches, and tunable optical antennas. △ Less

Submitted 28 November, 2018; v1 submitted 2 March, 2018; originally announced March 2018.

Comments: 35 pages, including supplementary information

Journal ref: Proc. Nat. Acad. Sci., 115, 12662-12667 (2018)

Carrier-Resolved Photo Hall Measurement in World-Record-Quality Perovskite and Kesterite Solar Absorbers

Authors: Oki Gunawan, Seong Ryul Pae, Douglas M. Bishop, Yun Seog Lee, Yudistira Virgus, Nam Joong Jeon, Jun Hong Noh, Xiaoyan Shao, Teodor Todorov, David B. Mitzi, Byungha Shin

Abstract: Majority and minority carrier properties such as type, density and mobility represent fundamental yet difficult to access parameters governing semiconductor device performance, most notably solar cells. Obtaining this information simultaneously under light illumination would unlock many critical parameters such as recombination lifetime, recombination coefficient, and diffusion length; while deepl… ▽ More Majority and minority carrier properties such as type, density and mobility represent fundamental yet difficult to access parameters governing semiconductor device performance, most notably solar cells. Obtaining this information simultaneously under light illumination would unlock many critical parameters such as recombination lifetime, recombination coefficient, and diffusion length; while deeply interesting for optoelectronic devices, this goal has remained elusive. We demonstrate here a new carrier-resolved photo-Hall technique that rests on a new identity relating hole-electron mobility difference ($Δμ$), Hall coefficient ($h$), and conductivity ($σ$): $Δμ=(2+d\ln h/d\ln σ)\,h\,σ$, and a rotating parallel dipole line ac-field Hall system with Fourier/lock-in detection for clean Hall signal measurement. We successfully apply this technique to recent world-record-quality perovskite and kesterite films and map the results against varying light intensities, demonstrating unprecedented simultaneous access to the above-mentioned parameters. △ Less

Submitted 22 February, 2018; originally announced February 2018.

Comments: 13 pages, 3 figures, supplementary materials (15 pages) including link to Movie S1

A Praise for Defensive Programming: Leveraging Uncertainty for Effective Malware Mitigation

Authors: Ruimin Sun, Marcus Botacin, Nikolaos Sapountzis, Xiaoyong Yuan, Matt Bishop, Donald E Porter, Xiaolin Li, Andre Gregio, Daniela Oliveira

Abstract: A promising avenue for improving the effectiveness of behavioral-based malware detectors would be to combine fast traditional machine learning detectors with high-accuracy, but time-consuming deep learning models. The main idea would be to place software receiving borderline classifications by traditional machine learning methods in an environment where uncertainty is added, while software is anal… ▽ More A promising avenue for improving the effectiveness of behavioral-based malware detectors would be to combine fast traditional machine learning detectors with high-accuracy, but time-consuming deep learning models. The main idea would be to place software receiving borderline classifications by traditional machine learning methods in an environment where uncertainty is added, while software is analyzed by more time-consuming deep learning models. The goal of uncertainty would be to rate-limit actions of potential malware during the time consuming deep analysis. In this paper, we present a detailed description of the analysis and implementation of CHAMELEON, a framework for realizing this uncertain environment for Linux. CHAMELEON offers two environments for software: (i) standard - for any software identified as benign by conventional machine learning methods and (ii) uncertain - for software receiving borderline classifications when analyzed by these conventional machine learning methods. The uncertain environment adds obstacles to software execution through random perturbations applied probabilistically on selected system calls. We evaluated CHAMELEON with 113 applications and 100 malware samples for Linux. Our results showed that at threshold 10%, intrusive and non-intrusive strategies caused approximately 65% of malware to fail accomplishing their tasks, while approximately 30% of the analyzed benign software to meet with various levels of disruption. With a dynamic, per-system call threshold, CHAMELEON caused 92% of the malware to fail, and only 10% of the benign software to be disrupted. We also found that I/O-bound software was three times more affected by uncertainty than CPU-bound software. Further, we analyzed the logs of software crashed with non-intrusive strategies, and found that some crashes are due to the software bugs. △ Less

Submitted 12 June, 2020; v1 submitted 7 February, 2018; originally announced February 2018.

Journal ref: IEEE Transaction of Dependability and Security 2020

Spectral gaps for the Two-Species Product Vacua and Boundary States models on the $d$-dimensional lattice

Authors: Michael Bishop

Abstract: We study the two-species Product Vacua and Boundary States (PVBS) models on the integer lattice $\mathbb{Z}^d$ and prove the existence and non-existence of a spectral gap for all choices of parameters. The PVBS models are spin-1 quantum spin systems which are translation-invariant, frustration-free, and composed of nearest-neighbor non-commuting interactions with both an exclusion property and an… ▽ More We study the two-species Product Vacua and Boundary States (PVBS) models on the integer lattice $\mathbb{Z}^d$ and prove the existence and non-existence of a spectral gap for all choices of parameters. The PVBS models are spin-1 quantum spin systems which are translation-invariant, frustration-free, and composed of nearest-neighbor non-commuting interactions with both an exclusion property and an interchange interaction between particle species. These models serve as possible representatives of families of automorphically equivalent gapped quantum spin-1 systems on $\mathbb{Z}^d$. The main result is that the two-species PVBS Hamiltonians have a positive spectral gap when gapped on both of the single-species subspaces and are gapless if gapless on either single-species subspace. The addition of a new particle species does not create any new gapless phases. △ Less

Submitted 12 March, 2018; v1 submitted 12 May, 2017; originally announced May 2017.

Comments: 38 pages

LeakSemantic: Identifying Abnormal Sensitive Network Transmissions in Mobile Applications

Authors: Hao Fu, Zizhan Zheng, Somdutta Bose, Matt Bishop, Prasant Mohapatra

Abstract: Mobile applications (apps) often transmit sensitive data through network with various intentions. Some transmissions are needed to fulfill the app's functionalities. However, transmissions with malicious receivers may lead to privacy leakage and tend to behave stealthily to evade detection. The problem is twofold: how does one unveil sensitive transmissions in mobile apps, and given a sensitive tr… ▽ More Mobile applications (apps) often transmit sensitive data through network with various intentions. Some transmissions are needed to fulfill the app's functionalities. However, transmissions with malicious receivers may lead to privacy leakage and tend to behave stealthily to evade detection. The problem is twofold: how does one unveil sensitive transmissions in mobile apps, and given a sensitive transmission, how does one determine if it is legitimate? In this paper, we propose LeakSemantic, a framework that can automatically locate abnormal sensitive network transmissions from mobile apps. LeakSemantic consists of a hybrid program analysis component and a machine learning component. Our program analysis component combines static analysis and dynamic analysis to precisely identify sensitive transmissions. Compared to existing taint analysis approaches, LeakSemantic achieves better accuracy with fewer false positives and is able to collect runtime data such as network traffic for each transmission. Based on features derived from the runtime data, machine learning classifiers are built to further differentiate between the legal and illegal disclosures. Experiments show that LeakSemantic achieves 91% accuracy on 2279 sensitive connections from 1404 apps. △ Less

Submitted 6 February, 2017; v1 submitted 3 February, 2017; originally announced February 2017.

ALMA Cycle 0 Publication Statistics

Authors: Felix Stoehr, Uta Grothkopf, Silvia Meakins, Marsha Bishop, Ayako Uchida, Leonardo Testi, Daisuke Iono, Kenichi Tatematsu, Al Wootten

Abstract: The scientific impact of a facility is the most important measure of its success. Monitoring and analysing the scientific return can help to modify and optimise operations and adapt to the changing needs of scientific research. The methodology that we have developed to monitor the scientific productivity of the ALMA Observatory, as well as the first results, are described. We focus on the outcome… ▽ More The scientific impact of a facility is the most important measure of its success. Monitoring and analysing the scientific return can help to modify and optimise operations and adapt to the changing needs of scientific research. The methodology that we have developed to monitor the scientific productivity of the ALMA Observatory, as well as the first results, are described. We focus on the outcome of the first cycle (Cycle 0) of ALMA Early Science operations. Despite the fact that only two years have passed since the completion of Cycle 0 and operations have already changed substantially, this analysisconfirms the effectiveness of the underlying concepts. We find that ALMA is fulfilling its promise as a transformational facility for the observation of the Universe in the submillimetre. △ Less

Submitted 18 January, 2016; originally announced January 2016.

Comments: 5 pages, 9 figures

Journal ref: The Messenger, 162, p. 30 (2015)

Spectral Gap and Edge Excitations of $d$-dimensional PVBS models on half-spaces

Authors: Michael Bishop, Bruno Nachtergaele, Amanda Young

Abstract: We analyze a class of quantum spin models defined on half-spaces in the $d$-dimensional hypercubic lattice bounded by a hyperplane with inward unit normal vector $m\in\mathbb{R}^d$. The family of models was previously introduced as the single species Product Vacua with Boundary States (PVBS) model, which is a spin-$1/2$ model with a XXZ-type nearest neighbor interactions depending on parameters… ▽ More We analyze a class of quantum spin models defined on half-spaces in the $d$-dimensional hypercubic lattice bounded by a hyperplane with inward unit normal vector $m\in\mathbb{R}^d$. The family of models was previously introduced as the single species Product Vacua with Boundary States (PVBS) model, which is a spin-$1/2$ model with a XXZ-type nearest neighbor interactions depending on parameters $λ_j\in (0,\infty)$, one for each coordinate direction. For any given values of the parameters, we prove an upper bound for the spectral gap above the unique ground state of these models, which vanishes for exactly one direction of the normal vector $m$. For all other choices of $m$ we derive a positive lower bound of the spectral gap, except for the case $λ_1 =\cdots =λ_d=1$, which is known to have gapless excitations in the bulk. △ Less

Submitted 24 September, 2015; originally announced September 2015.

Journal ref: J. Stat. Phys. 162 (2016) 1485--1521

Coarsening dynamics of binary liquids with active rotation

Authors: Syeda Sabrina, Matthew Spellings, Sharon C. Glotzer, Kyle J. M. Bishop

Abstract: Active matter comprised of many self-driven units can exhibit emergent collective behaviors such as pattern formation and phase separation in both biologica and synthetic systems. While these behaviors are increasingly well understood for ensembles of linearly self-propelled particles, less is known about the collective behaviors of active rotating particles where energy input at the particle leve… ▽ More Active matter comprised of many self-driven units can exhibit emergent collective behaviors such as pattern formation and phase separation in both biologica and synthetic systems. While these behaviors are increasingly well understood for ensembles of linearly self-propelled particles, less is known about the collective behaviors of active rotating particles where energy input at the particle level gives rise to rotational particle motion. A recent simulation study revealed that active rotation can induce phase separation in mixtures of counter-rotating particles in 2D. In contrast to that of linearly self-propelled particles, the phase separation of counter-rotating fluids is accompanied by steady convective flows that originate at the fluid-fluid interface. Here, we investigate the influence of these flows on the coarsening dynamics of actively rotating binary liquids using a phenomenological, hydrodynamic model that combines a Cahn-Hilliard equation for the fluid composition with a Navier-Stokes equation for the fluid velocity. The effect of active rotation is introduced though an additional force within the Navier-Stokes equations that arises due to gradients in the concentrations of clockwise and counter-clockwise rotating particles. Depending on the strength of active rotation and that of frictional interactions with the stationary surroundings, we observe and explain new dynamical behaviors such as "active coarsening" via self-generated flows as well as the emergence of self-propelled vortex doublets. We confirm that many of the qualitative behaviors identified by the continuum model can also be found in discrete, particle-based simulations of actively rotating liquids. Our results highlight further opportunities for achieving complex dissipative structures in active materials subject to distributed actuation. △ Less

Submitted 23 July, 2015; originally announced July 2015.

High Photoresponsivity and Short Photo Response Times in Few-Layered WSe$_2$ Transistors

Authors: Nihar R. Pradhan, Jonathan Ludwig, Zhengguang Lu, Daniel Rhodes, Michael M. Bishop, Komalavalli Thirunavukkuarasu, Stephen A. McGill, Dmitry Smirnov, Luis Balicas

Abstract: Here, we report the photoconducting response of field-effect transistors based on three atomic layers of chemical vapor transport grown WSe$_2$ crystals mechanically exfoliated onto SiO$_2$. We find that tri-layered WSe$_2$ field-effect transistors, built with the simplest possible architecture, can display high hole mobilities ranging from 350 cm$^2$/Vs at room temperature (saturating at a value… ▽ More Here, we report the photoconducting response of field-effect transistors based on three atomic layers of chemical vapor transport grown WSe$_2$ crystals mechanically exfoliated onto SiO$_2$. We find that tri-layered WSe$_2$ field-effect transistors, built with the simplest possible architecture, can display high hole mobilities ranging from 350 cm$^2$/Vs at room temperature (saturating at a value of ~500 cm$^2$/Vs below 50 K) displaying a strong photocurrent response which leads to exceptionally high photo responsivities up to 7 A/W under white light illumination of the entire channel for power densities p < 10$^2$ W/m$^2$. Under a fixed wavelength of $λ$ = 532 nm and a laser spot size smaller than the conducting channel area we extract photo responsitivities approaching 100 mA/W with concomitantly high external quantum efficiencies up to ~ 40 % at room temperature. These values surpass values recently reported from more complex architectures, such as graphene and transition metal dichalcogenides based heterostructures. Also, tri-layered WSe$_2$ photo-transistors display photo response times in the order of 10 microseconds. Our results indicate that the addition of a few atomic layers considerably decreases the photo response times, probably by minimizing the interaction with the substrates, while maintaining a very high photo-responsivity. △ Less

Submitted 19 May, 2015; originally announced May 2015.

Comments: 39 pages, 6 figures, and supplemental information including 5 supplemental figures. To appear in ACS Applied Materials and Interfaces

Journal ref: ACS Applied Materials & Interfaces, 7, 22, 12080 (2015)

Two years of ALMA bibliography - lessons learned

Authors: Silvia Meakins, Uta Grothkopf, Marsha J. Bishop, Felix Stoehr, Ken Tatematsu

Abstract: Telescope bibliographies are integral parts of observing facilities. They are used to associate the published literature with archived observational data, to measure an observatory's scientific output through publication and citation statistics, and to define guidelines for future observing strategies. The ESO and NRAO librarians as well as NAOJ jointly maintain the ALMA (Atacama Large Millimete… ▽ More Telescope bibliographies are integral parts of observing facilities. They are used to associate the published literature with archived observational data, to measure an observatory's scientific output through publication and citation statistics, and to define guidelines for future observing strategies. The ESO and NRAO librarians as well as NAOJ jointly maintain the ALMA (Atacama Large Millimeter/submillimeter Array) bibliography, a database of refereed papers that use ALMA data. In this paper, we illustrate how relevant articles are identified, which procedures are used to tag entries in the database and link them to the correct observations, and how results are communicated to ALMA stakeholders and the wider community. Efforts made to streamline the process will be explained and evaluated, and a first analysis of ALMA papers published after two years of observations will be given. △ Less

Submitted 25 July, 2014; originally announced July 2014.

Comments: 7 pages; to be published in the Proceedings of SPIE, vol. 9149, 9149-81 (2014)

Computations for Coxeter arrangements and Solomon's descent algebra III: Groups of rank seven and eight

Authors: Marcus Bishop, J. Matthew Douglass, Goetz Pfeiffer, Gerhard Roehrle

Abstract: In this paper we extend the computations in parts I and II of this series of papers and complete the proof of a conjecture of Lehrer and Solomon expressing the character of a finite Coxeter group W acting on the pth graded component of its Orlik-Solomon algebra as a sum of characters induced from linear characters of centralizers of elements of W for groups of rank seven and eight. For classical C… ▽ More In this paper we extend the computations in parts I and II of this series of papers and complete the proof of a conjecture of Lehrer and Solomon expressing the character of a finite Coxeter group W acting on the pth graded component of its Orlik-Solomon algebra as a sum of characters induced from linear characters of centralizers of elements of W for groups of rank seven and eight. For classical Coxeter groups, these characters are given using a formula that is expected to hold in all ranks. △ Less

Submitted 15 November, 2014; v1 submitted 25 March, 2014; originally announced March 2014.

Comments: Minor changes; final version

MSC Class: 20F55; 20C40; 20C15

Lifschitz Tails for Random Schrödinger Operator in Bernoulli Distributed Potentials

Authors: Michael Bishop, Vita Borovyk, Jan Wehr

Abstract: This paper presents an elementary proof of Lifschitz tail behavior for random discrete Schrödinger operators with a Bernoulli-distributed potential. The proof approximates the low eigenvalues by eigenvalues of sine waves supported where the potential takes its lower value. This is motivated by the idea that the eigenvectors associated to the low eigenvalues react to the jump in the values of the p… ▽ More This paper presents an elementary proof of Lifschitz tail behavior for random discrete Schrödinger operators with a Bernoulli-distributed potential. The proof approximates the low eigenvalues by eigenvalues of sine waves supported where the potential takes its lower value. This is motivated by the idea that the eigenvectors associated to the low eigenvalues react to the jump in the values of the potential as if the gap were infinite. △ Less

Submitted 21 March, 2014; originally announced March 2014.

On the Quiver Presentation of the Descent Algebra of the Hyperoctahedral Group

Authors: Marcus Bishop

Abstract: In a recent article we introduced a mechanism for producing a presentation of the descent algebra of the symmetric group as a quiver with relations, the mechanism arising from a new construction of the descent algebra as a homomorphic image of an algebra of binary forests. Here we extend the method to construct a similar presentation of the descent algebra of the hyperoctahedral group, providing a… ▽ More In a recent article we introduced a mechanism for producing a presentation of the descent algebra of the symmetric group as a quiver with relations, the mechanism arising from a new construction of the descent algebra as a homomorphic image of an algebra of binary forests. Here we extend the method to construct a similar presentation of the descent algebra of the hyperoctahedral group, providing a simple proof of the known formula for the quiver of this algebra and a straightforward method for calculating the relations. △ Less

Submitted 11 August, 2014; v1 submitted 12 July, 2013; originally announced July 2013.

MSC Class: 16G20

Mixture Representations for Inference and Learning in Boltzmann Machines

Authors: Neil D. Lawrence, Christopher M. Bishop, Michael I. Jordan

Abstract: Boltzmann machines are undirected graphical models with two-state stochastic variables, in which the logarithms of the clique potentials are quadratic functions of the node states. They have been widely studied in the neural computing literature, although their practical applicability has been limited by the difficulty of finding an effective learning algorithm. One well-established approach, know… ▽ More Boltzmann machines are undirected graphical models with two-state stochastic variables, in which the logarithms of the clique potentials are quadratic functions of the node states. They have been widely studied in the neural computing literature, although their practical applicability has been limited by the difficulty of finding an effective learning algorithm. One well-established approach, known as mean field theory, represents the stochastic distribution using a factorized approximation. However, the corresponding learning algorithm often fails to find a good solution. We conjecture that this is due to the implicit uni-modality of the mean field approximation which is therefore unable to capture multi-modality in the true distribution. In this paper we use variational methods to approximate the stochastic distribution using multi-modal mixtures of factorized distributions. We present results for both inference and learning to demonstrate the effectiveness of this approach. △ Less

Submitted 30 January, 2013; originally announced January 2013.

Comments: Appears in Proceedings of the Fourteenth Conference on Uncertainty in Artificial Intelligence (UAI1998)

Report number: UAI-P-1998-PG-320-327

Variational Relevance Vector Machines

Authors: Christopher M. Bishop, Michael Tip**

Abstract: The Support Vector Machine (SVM) of Vapnik (1998) has become widely established as one of the leading approaches to pattern recognition and machine learning. It expresses predictions in terms of a linear combination of kernel functions centred on a subset of the training data, known as support vectors. Despite its widespread success, the SVM suffers from some important limitations, one of the mos… ▽ More The Support Vector Machine (SVM) of Vapnik (1998) has become widely established as one of the leading approaches to pattern recognition and machine learning. It expresses predictions in terms of a linear combination of kernel functions centred on a subset of the training data, known as support vectors. Despite its widespread success, the SVM suffers from some important limitations, one of the most significant being that it makes point predictions rather than generating predictive distributions. Recently Tip** (1999) has formulated the Relevance Vector Machine (RVM), a probabilistic model whose functional form is equivalent to the SVM. It achieves comparable recognition accuracy to the SVM, yet provides a full predictive distribution, and also requires substantially fewer kernel functions. The original treatment of the RVM relied on the use of type II maximum likelihood (the `evidence framework') to provide point estimates of the hyperparameters which govern model sparsity. In this paper we show how the RVM can be formulated and solved within a completely Bayesian paradigm through the use of variational inference, thereby giving a posterior distribution over both parameters and hyperparameters. We demonstrate the practicality and performance of the variational RVM using both synthetic and real world examples. △ Less

Submitted 16 January, 2013; originally announced January 2013.

Comments: Appears in Proceedings of the Sixteenth Conference on Uncertainty in Artificial Intelligence (UAI2000)

Report number: UAI-P-2000-PG-46-53

Bayesian Hierarchical Mixtures of Experts

Authors: Christopher M. Bishop, Markus Svensen

Abstract: The Hierarchical Mixture of Experts (HME) is a well-known tree-based model for regression and classification, based on soft probabilistic splits. In its original formulation it was trained by maximum likelihood, and is therefore prone to over-fitting. Furthermore the maximum likelihood framework offers no natural metric for optimizing the complexity and structure of the tree. Previous attempts to… ▽ More The Hierarchical Mixture of Experts (HME) is a well-known tree-based model for regression and classification, based on soft probabilistic splits. In its original formulation it was trained by maximum likelihood, and is therefore prone to over-fitting. Furthermore the maximum likelihood framework offers no natural metric for optimizing the complexity and structure of the tree. Previous attempts to provide a Bayesian treatment of the HME model have relied either on ad-hoc local Gaussian approximations or have dealt with related models representing the joint distribution of both input and output variables. In this paper we describe a fully Bayesian treatment of the HME model based on variational inference. By combining local and global variational methods we obtain a rigourous lower bound on the marginal probability of the data under the model. This bound is optimized during the training phase, and its resulting value can be used for model order selection. We present results using this approach for a data set describing robot arm kinematics. △ Less

Submitted 19 October, 2012; originally announced December 2012.

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

Report number: UAI-P-2003-PG-57-64

Ground State Energy of Mean-field Model of Interacting Bosons in Bernoulli Potential

Authors: Michael Bishop, Jan Wehr

Abstract: This paper explores a system of interacting `soft core' bosons in the Gross-Pitaevskii mean-field approximation in a random Bernoulli potential. First, a condition for delocalization of the ground state wave function is proved which depends on the number of particles and interaction strength. Using this condition, asymptotics for ground state energy per particle are derived in the large system lim… ▽ More This paper explores a system of interacting `soft core' bosons in the Gross-Pitaevskii mean-field approximation in a random Bernoulli potential. First, a condition for delocalization of the ground state wave function is proved which depends on the number of particles and interaction strength. Using this condition, asymptotics for ground state energy per particle are derived in the large system limit for small values of the coupling constant. Our methods directly describe the shape of the ground state in a given realization of the random potential. △ Less

Submitted 27 December, 2012; v1 submitted 6 December, 2012; originally announced December 2012.

Comments: 18 pages

MSC Class: 81V70; 35Q55; 47J20; 82B44; 47H40; 35A15

On the Quiver Presentation of the Descent Algebra of the Symmetric Group

Authors: Marcus Bishop, Götz Pfeiffer

Abstract: We describe a presentation for the descent algebra of the symmetric group $\sym{n}$ as a quiver with relations. This presentation arises from a new construction of the descent algebra as a homomorphic image of an algebra of forests of binary trees which can be identified with a subspace of the free Lie algebra. In this setting, we provide a new short proof of the known fact that the quiver of the… ▽ More We describe a presentation for the descent algebra of the symmetric group $\sym{n}$ as a quiver with relations. This presentation arises from a new construction of the descent algebra as a homomorphic image of an algebra of forests of binary trees which can be identified with a subspace of the free Lie algebra. In this setting, we provide a new short proof of the known fact that the quiver of the descent algebra of $\sym{n}$ is given by restricted partition refinement. Moreover, we describe certain families of relations and conjecture that for fixed $n\in\mathbb{N}$, the finite set of relations from these families that are relevant for the descent algebra of $\sym{n}$ generates the ideal of relations, and hence yields an explicit presentation by generators and relations of the algebra. △ Less

Submitted 25 March, 2013; v1 submitted 1 June, 2012; originally announced June 2012.

Comments: Final version; 21 pages

MSC Class: 16G20; 20F55