-
A Pattern Language for Machine Learning Tasks
Authors:
Benjamin Rodatz,
Ian Fan,
Tuomas Laakkonen,
Neil John Ortega,
Thomas Hoffman,
Vincent Wang-Mascianica
Abstract:
Idealised as universal approximators, learners such as neural networks can be viewed as "variable functions" that may become one of a range of concrete functions after training. In the same way that equations constrain the possible values of variables in algebra, we may view objective functions as constraints on the behaviour of learners. We extract the equivalences perfectly optimised objective f…
▽ More
Idealised as universal approximators, learners such as neural networks can be viewed as "variable functions" that may become one of a range of concrete functions after training. In the same way that equations constrain the possible values of variables in algebra, we may view objective functions as constraints on the behaviour of learners. We extract the equivalences perfectly optimised objective functions impose, calling them "tasks". For these tasks, we develop a formal graphical language that allows us to: (1) separate the core tasks of a behaviour from its implementation details; (2) reason about and design behaviours model-agnostically; and (3) simply describe and unify approaches in machine learning across domains.
As proof-of-concept, we design a novel task that enables converting classifiers into generative models we call "manipulators", which we implement by directly translating task specifications into code. The resulting models exhibit capabilities such as style transfer and interpretable latent-space editing, without the need for custom architectures, adversarial training or random sampling. We formally relate the behaviour of manipulators to GANs, and empirically demonstrate their competitive performance with VAEs. We report on experiments across vision and language domains aiming to characterise manipulators as approximate Bayesian inversions of discriminative classifiers.
△ Less
Submitted 2 July, 2024;
originally announced July 2024.
-
Comparing Optimization Targets for Contrast-Consistent Search
Authors:
Hugo Fry,
Seamus Fallows,
Ian Fan,
Jamie Wright,
Nandi Schoots
Abstract:
We investigate the optimization target of Contrast-Consistent Search (CCS), which aims to recover the internal representations of truth of a large language model. We present a new loss function that we call the Midpoint-Displacement (MD) loss function. We demonstrate that for a certain hyper-parameter value this MD loss function leads to a prober with very similar weights to CCS. We further show t…
▽ More
We investigate the optimization target of Contrast-Consistent Search (CCS), which aims to recover the internal representations of truth of a large language model. We present a new loss function that we call the Midpoint-Displacement (MD) loss function. We demonstrate that for a certain hyper-parameter value this MD loss function leads to a prober with very similar weights to CCS. We further show that this hyper-parameter is not optimal and that with a better hyper-parameter the MD loss function attains a higher test accuracy than CCS.
△ Less
Submitted 1 November, 2023;
originally announced November 2023.
-
Aircraft Skin Inspections: Towards a New Model for Dent Evaluation
Authors:
Pasquale Lafiosca,
Ip-Shing Fan,
Nicolas P. Avdelidis
Abstract:
Aircraft maintenance, repair and overhaul industry is gradually switching to 3D scanning for dent inspection. High-accuracy devices allow quick and repeatable measurements, which translate into efficient reporting and more objective damage evaluations. However, the potential of 3D scanners is far from being exploited. This is due to the traditional way in which the structural repair manual deals w…
▽ More
Aircraft maintenance, repair and overhaul industry is gradually switching to 3D scanning for dent inspection. High-accuracy devices allow quick and repeatable measurements, which translate into efficient reporting and more objective damage evaluations. However, the potential of 3D scanners is far from being exploited. This is due to the traditional way in which the structural repair manual deals with dents, that is, considering length, width and depth as the only relevant measures. Being equivalent to describing a dent similarly to a box, the current approach discards any information about the actual shape. This causes high degrees of ambiguity, with very different shapes (and corresponding fatigue life) being classified as the same, and nullifies the effort of acquiring such great amount of information from high-accuracy 3D scanners. In this paper a 7-parameter model is proposed to describe the actual dent shape, thus enabling the exploitation of the high fidelity data produced by 3D scanners. The compact set of values can then be compared against historical data and structural evaluations based on the same model. The proposed approach has been evaluated in both simulations and point cloud data generated by 8tree's dentCHECK tool, suggesting increased capability to evaluate damage, enabling more targeted interventions and, ultimately, saving costs.
△ Less
Submitted 11 July, 2023; v1 submitted 25 January, 2023;
originally announced January 2023.
-
Automatic Segmentation of Aircraft Dents in Point Clouds
Authors:
Pasquale Lafiosca,
Ip-Shing Fan,
Nicolas P. Avdelidis
Abstract:
Dents on the aircraft skin are frequent and may easily go undetected during airworthiness checks, as their inspection process is tedious and extremely subject to human factors and environmental conditions. Nowadays, 3D scanning technologies are being proposed for more reliable, human-independent measurements, yet the process of inspection and reporting remains laborious and time consuming because…
▽ More
Dents on the aircraft skin are frequent and may easily go undetected during airworthiness checks, as their inspection process is tedious and extremely subject to human factors and environmental conditions. Nowadays, 3D scanning technologies are being proposed for more reliable, human-independent measurements, yet the process of inspection and reporting remains laborious and time consuming because data acquisition and validation are still carried out by the engineer. For full automation of dent inspection, the acquired point cloud data must be analysed via a reliable segmentation algorithm, releasing humans from the search and evaluation of damage. This paper reports on two developments towards automated dent inspection. The first is a method to generate a synthetic dataset of dented surfaces to train a fully convolutional neural network. The training of machine learning algorithms needs a substantial volume of dent data, which is not readily available. Dents are thus simulated in random positions and shapes, within criteria and definitions of a Boeing 737 structural repair manual. The noise distribution from the scanning apparatus is then added to reflect the complete process of 3D point acquisition on the training. The second proposition is a surface fitting strategy to convert 3D point clouds to 2.5D. This allows higher resolution point clouds to be processed with a small amount of memory compared with state-of-the-art methods involving 3D sampling approaches. Simulations with available ground truth data show that the proposed technique reaches an intersection-over-union of over 80%. Experiments over dent samples prove an effective detection of dents with a speed of over 500 000 points per second.
△ Less
Submitted 24 February, 2023; v1 submitted 3 May, 2022;
originally announced May 2022.
-
lambeq: An Efficient High-Level Python Library for Quantum NLP
Authors:
Dimitri Kartsaklis,
Ian Fan,
Richie Yeung,
Anna Pearson,
Robin Lorenz,
Alexis Toumi,
Giovanni de Felice,
Konstantinos Meichanetzidis,
Stephen Clark,
Bob Coecke
Abstract:
We present lambeq, the first high-level Python library for Quantum Natural Language Processing (QNLP). The open-source toolkit offers a detailed hierarchy of modules and classes implementing all stages of a pipeline for converting sentences to string diagrams, tensor networks, and quantum circuits ready to be used on a quantum computer. lambeq supports syntactic parsing, rewriting and simplificati…
▽ More
We present lambeq, the first high-level Python library for Quantum Natural Language Processing (QNLP). The open-source toolkit offers a detailed hierarchy of modules and classes implementing all stages of a pipeline for converting sentences to string diagrams, tensor networks, and quantum circuits ready to be used on a quantum computer. lambeq supports syntactic parsing, rewriting and simplification of string diagrams, ansatz creation and manipulation, as well as a number of compositional models for preparing quantum-friendly representations of sentences, employing various degrees of syntax sensitivity. We present the generic architecture and describe the most important modules in detail, demonstrating the usage with illustrative examples. Further, we test the toolkit in practice by using it to perform a number of experiments on simple NLP tasks, implementing both classical and quantum pipelines.
△ Less
Submitted 8 October, 2021;
originally announced October 2021.
-
Revisiting $^{129}$Xe electric dipole moment measurements applying a new global phase fitting approach
Authors:
T. Liu,
K. Rolfs,
I. Fan,
S. Haude,
W. Kilian,
L. Li,
A. Schnabel,
J. Voigt,
L. Trahms
Abstract:
By measuring the nuclear magnetic spin precession frequencies of polarized $^{129}$Xe and $^{3}$He, a new upper limit on the $^{129}$Xe atomic electric dipole moment (EDM) $ d_\mathrm{A} (^{129}\mathrm{Xe})$ was reported in Phys. Rev. Lett. 123, 143003 (2019). Here, we propose a new evaluation method based on global phase fitting (GPF) for analyzing the continuous phase development of the $^{3}$He…
▽ More
By measuring the nuclear magnetic spin precession frequencies of polarized $^{129}$Xe and $^{3}$He, a new upper limit on the $^{129}$Xe atomic electric dipole moment (EDM) $ d_\mathrm{A} (^{129}\mathrm{Xe})$ was reported in Phys. Rev. Lett. 123, 143003 (2019). Here, we propose a new evaluation method based on global phase fitting (GPF) for analyzing the continuous phase development of the $^{3}$He-$^{129}$Xe comagnetometer signal. The Cramer-Rao Lower Bound on the $^{129}$Xe EDM for the GPF method is theoretically derived and shows the potential benefit of our new approach. The robustness of the GPF method is verified with Monte-Carlo studies. By optimizing the analysis parameters and adding data that could not be analyzed with the former method, we obtain a result of $d_\mathrm{A} (^{129}\mathrm{Xe}) = 1.1 \pm 3.6~\mathrm{(stat)} \pm 2.0~\mathrm{(syst)} \times 10^{-28}~ e~\mathrm{cm}$ in an unblinded analysis. For the systematic uncertainty analyses, we adopted all methods from the aforementioned PRL publication except the comagnetometer phase drift, which can be omitted using the GPF method. The updated null result can be interpreted as a new upper limit of $| d_\mathrm{A} (^{129}\mathrm{Xe}) | < 8.3 \times 10^{-28}~e~\mathrm{cm}$ at the 95\% C.L.
△ Less
Submitted 15 November, 2020; v1 submitted 18 August, 2020;
originally announced August 2020.
-
New Limit on the Permanent Electric Dipole Moment of $^{129}$Xe using $^{3}$He Comagnetometry and SQUID Detection
Authors:
N. Sachdeva,
I. Fan,
E. Babcock,
M. Burghoff,
T. E. Chupp,
S. Degenkolb,
P. Fierlinger,
S. Haude,
E. Kraegeloh,
W. Kilian,
S. Knappe-Grüneberg,
F. Kuchler,
T. Liu,
M. Marino,
J. Meinel,
K. Rolfs,
Z. Salhi,
A. Schnabel,
J. T. Singh,
S. Stuiber,
W. A. Terrano,
L. Trahms,
J. Voigt
Abstract:
We report results of a new technique to measure the electric dipole moment of $^{129}$Xe with $^3$He comagnetometry. Both species are polarized using spin-exchange optical pum**, transferred to a measurement cell, and transported into a magnetically shielded room, where SQUID magnetometers detect free precession in applied electric and magnetic fields. The result from a one week measurement camp…
▽ More
We report results of a new technique to measure the electric dipole moment of $^{129}$Xe with $^3$He comagnetometry. Both species are polarized using spin-exchange optical pum**, transferred to a measurement cell, and transported into a magnetically shielded room, where SQUID magnetometers detect free precession in applied electric and magnetic fields. The result from a one week measurement campaign in 2017 and a 2.5 week campaign in 2018, combined with detailed study of systematic effects, is $d_A(^{129}\mathrm{Xe}) = (1.4 \pm 6.6_\mathrm{stat} \pm 2.0_\mathrm{syst})\times10^{-28}~e\,\mathrm{cm}$. This corresponds to an upper limit of $|d_A(^{129}\mathrm{Xe})| < 1.4 \times 10^{-27} ~e\,\mathrm{cm}~(95\%~\mathrm{CL})$, a factor of five more sensitive than the limit set in 2001.
△ Less
Submitted 26 September, 2019;
originally announced September 2019.
-
A new measurement of the permanent electric dipole moment of $^{129}$Xe using $^{3}$He comagnetometry and SQUID detection
Authors:
N. Sachdeva,
I. Fan,
E. Babcock,
M. Burghoff,
T. E. Chupp,
S. Degenkolb,
P. Fierlinger,
E. Kraegeloh,
W. Kilian,
S. Knappe-Grüneberg,
F. Kuchler,
T. Liu,
M. Marino,
J. Meinel,
Z. Salhi,
A. Schnabel,
J. T. Singh,
S. Stuiber,
W. A. Terrano,
L. Trahms,
J. Voigt
Abstract:
We describe a new technique to measure the EDM of $^{129}$Xe with $^3$He comagnetometry. Both species are polarized using spin-exchange optical pum**, transferred to a measurement cell, and transported into a magnetically shielded room, where SQUID magnetometers detect free precession in applied electric and magnetic fields. The result of a one week run combined with a detailed study of systemat…
▽ More
We describe a new technique to measure the EDM of $^{129}$Xe with $^3$He comagnetometry. Both species are polarized using spin-exchange optical pum**, transferred to a measurement cell, and transported into a magnetically shielded room, where SQUID magnetometers detect free precession in applied electric and magnetic fields. The result of a one week run combined with a detailed study of systematic effects is $d_A(^{129}\mathrm{Xe}) = (0.26 \pm 2.33_\mathrm{stat} \pm 0.72_\mathrm{syst})\times10^{-27}~e\,\mathrm{cm}$. This corresponds to an upper limit of $|d_A(^{129}\mathrm{Xe})| < 4.81\times 10^{-27} ~e\,\mathrm{cm}~(95\%~\mathrm{CL})$, a factor of 1.4 more sensitive than the previous limit.
△ Less
Submitted 7 February, 2019;
originally announced February 2019.
-
A magnetically shielded room with ultra low residual field and gradient
Authors:
I. Altarev,
E. Babcock,
D. Beck,
M. Burghoff,
S. Chesnevskaya,
T. Chupp,
S. Degenkolb,
I. Fan,
P. Fierlinger,
A. Frei,
E. Gutsmiedl,
S. Knappe-Grüneberg,
F. Kuchler,
T. Lauer,
P. Link,
T. Lins,
M. Marino,
J. McAndrew,
B. Niessen,
S. Paul,
G. Petzoldt,
U. Schläpfer,
A. Schnabel,
S. Sharma,
J. Singh
, et al. (7 additional authors not shown)
Abstract:
A versatile and portable magnetically shielded room with a field of (700 \pm 200) pT within a central volume of 1m x 1m x 1m and a field gradient less than 300 pT/m is described. This performance represents more than a hundred-fold improvement of the state of the art for a two-layer magnetic shield and provides an environment suitable for a next generation of precision experiments in fundamental p…
▽ More
A versatile and portable magnetically shielded room with a field of (700 \pm 200) pT within a central volume of 1m x 1m x 1m and a field gradient less than 300 pT/m is described. This performance represents more than a hundred-fold improvement of the state of the art for a two-layer magnetic shield and provides an environment suitable for a next generation of precision experiments in fundamental physics at low energies; in particular, searches for electric dipole moments of fundamental systems and tests of Lorentz-invariance based on spin-precession experiments. Studies of the residual fields and their sources enable improved design of future ultra-low gradient environments and experimental apparatus.
△ Less
Submitted 24 March, 2014;
originally announced March 2014.
-
New Determination of Muonium-Deuterium 1S-2S Isotope Shift
Authors:
Isaac Fan,
Chun-Yu Chang,
Li-Bang Wang,
Simon L. Cornish,
Jow-Tsong Shy,
Yi-Wei Liu
Abstract:
We report a new determination of muonium 1S-2S transition frequency and its isotope shift with deuterium by recalibrating the iodine reference lines using an optical frequency comb. The reference lines for the muonium and deuterium 1S-2S transitions are determined with a precision of 2.4*10^-10 and 1.7*10^-10 respectively. A new muonium-deuterium 1S-2S isotope-shift frequency is derived from these…
▽ More
We report a new determination of muonium 1S-2S transition frequency and its isotope shift with deuterium by recalibrating the iodine reference lines using an optical frequency comb. The reference lines for the muonium and deuterium 1S-2S transitions are determined with a precision of 2.4*10^-10 and 1.7*10^-10 respectively. A new muonium-deuterium 1S-2S isotope-shift frequency is derived from these references to be 11 203 464.9(9.2)(4.0) MHz, in agreement with an updated bound-state quantum-electrodynamics prediction based on 2010 adjustments of Committee on Data for Science and Technology and 2.3 times better in the systematic uncertainty than the previous best determination.
△ Less
Submitted 6 October, 2013;
originally announced October 2013.
-
$β$ NMR study of Isolated $^8$Li in the enhanced paramagnet Platinum
Authors:
Oren Ofer,
K. H. Chow,
I. Fan,
M. Egilmez,
T. J. Parolin,
M. D. Hossain,
J. Jung,
Z. Salman,
R. F. Kiefl,
C. D. P. Levy,
G. D. Morris,
M. R. Pearson,
H. Saadaoui,
Q. Song,
D. Wang,
W. A. MacFarlane
Abstract:
We report β detected nuclear magnetic resonance (βNMR) measurements of 8Li+ implanted into high purity Pt. The frequency of the 8Li βNMR resonance and the spin-lattice relaxation rates 1/T1 were measured at temperatures ranging from 3 to 300 K. Remarkably, both the spin-lattice relaxation rate and the Knight shift K depend linearly on temperature T although the bulk susceptibility does not. K is f…
▽ More
We report β detected nuclear magnetic resonance (βNMR) measurements of 8Li+ implanted into high purity Pt. The frequency of the 8Li βNMR resonance and the spin-lattice relaxation rates 1/T1 were measured at temperatures ranging from 3 to 300 K. Remarkably, both the spin-lattice relaxation rate and the Knight shift K depend linearly on temperature T although the bulk susceptibility does not. K is found to scale with the Curie-Weiss dependence of the Pt susceptibility extrapolated to low temperatures. This is attributed to a defect response of the enhanced paramagnetism of Pt, i.e. the presence of the interstitial Li+ locally relieves the tendency for the Curie-Weiss susceptibility to saturate at low T . We propose that the low temperature saturation in \c{hi} of Pt may be related to an interband coupling between the s and d bands that is disrupted locally by the presence of the Li+.
△ Less
Submitted 19 August, 2012;
originally announced August 2012.
-
Absolute Frequency Measurement of the 378 nm Transition in Thallium
Authors:
Tzu-Ling Chen,
Isaac Fan,
Chang-Yi Lin,
Hsuan-Chen Chen,
Shih-En Chen,
Jow-Tsong Shy,
Yi-Wei Liu
Abstract:
The absolute frequency measurement for all of the hyperfine transitions of 6P1/2 -> 7S1/2 using a self-referencing frequency comb is reported. This transition can be used as a bench marker for the accurate atomic wave function, and then improve the calculation of the atomic parity- non-conservation (PNC). The frequency center is precisely determined by saturation spectroscopy utilizing a pair of t…
▽ More
The absolute frequency measurement for all of the hyperfine transitions of 6P1/2 -> 7S1/2 using a self-referencing frequency comb is reported. This transition can be used as a bench marker for the accurate atomic wave function, and then improve the calculation of the atomic parity- non-conservation (PNC). The frequency center is precisely determined by saturation spectroscopy utilizing a pair of the counter-propagating laser beams intersecting with atomic beam. Such a Doppler-free profile has been improved to 350 kHz in the absolute frequency measurement. The 7S1/2 state hyperfine splittings have also been deduced from our results.
△ Less
Submitted 20 March, 2013; v1 submitted 7 March, 2007;
originally announced March 2007.
-
$β$-NMR of Isolated $^{8}$Li$^{+}$ Implanted into a Thin Copper Film
Authors:
Z. Salman,
A. I. Mansour,
K. H. Chow,
M. Beaudoin,
I. Fan,
J. Jung,
T. A. Keeler,
R. F. Kiefl,
C. D. P. Levy,
R. C. Ma,
G. D. Morris,
T. J. Parolin,
D. Wang,
W. A. MacFarlane
Abstract:
Depth-controlled $β$-NMR was used to study highly spin-polarized $^8$Li in a Cu film of thickness 100 nm deposited onto a MgO substrate. The positive Knight Shifts and spin relaxation data show that $^8$Li occupies two sites at low temperatures, assigned to be the substitutional ($S$) and octahedral ($O$) interstitial sites. Between 50 to 100 K, there is a site change from $O$ to $S$. The temper…
▽ More
Depth-controlled $β$-NMR was used to study highly spin-polarized $^8$Li in a Cu film of thickness 100 nm deposited onto a MgO substrate. The positive Knight Shifts and spin relaxation data show that $^8$Li occupies two sites at low temperatures, assigned to be the substitutional ($S$) and octahedral ($O$) interstitial sites. Between 50 to 100 K, there is a site change from $O$ to $S$. The temperature dependence of the Knight shifts and spin-lattice relaxation rates at high temperatures, i.e. when all the Li are in the $S$ site, is consistent with the Korringa Law for a simple metal.
△ Less
Submitted 29 December, 2006;
originally announced January 2007.
