Showing 1–32 of 32 results for author: Chrzanowski, M

Noise analysis of a quasi-phase-matched quantum frequency converter and higher-order counter-propagating SPDC

Authors: Felix Mann, Helen M. Chrzanowski, Felipe Gewers, Marlon Placke, Sven Ramelow

Abstract: Quantum frequency conversion (QFC) will be an indispensable ingredient in future quantum technologies. For example, large-scale fibre-based quantum networks will require QFC to interconnect heterogeneous building blocks like emitters, channels, memories and detectors. The performance of existing QFC devices - typically realised in periodically-poled nonlinear crystals - is often severely limited b… ▽ More Quantum frequency conversion (QFC) will be an indispensable ingredient in future quantum technologies. For example, large-scale fibre-based quantum networks will require QFC to interconnect heterogeneous building blocks like emitters, channels, memories and detectors. The performance of existing QFC devices - typically realised in periodically-poled nonlinear crystals - is often severely limited by parasitic noise that arises when the pump wavelength lies between the wavelengths which are inter-converted. Here we investigate the noise spectrum of a frequency converter pumped by a CW 1064 nm laser. The converter was realised as a monolithic bulk pump enhancement cavity made from a periodically-poled potassium titanyl phosphate (ppKTP) crystal - quasi-phase-matched for the conversion of 637 nm to 1587 nm. In the range from 1140 nm to 1330 nm (up to 60 THz from the pump) Stokes-Raman resonances can be identified as the dominant noise source while the noise in the range from 1330 nm to 1650 nm can be attributed mainly to parasitic spontaneous parametric down-conversion (SPDC). Further, a succession of narrow-band peaks is observed in the spectrum originating from higher-order counter-propagating SPDC. Both types of counter-propagation, where either the lower-energy idler photon or the higher-energy signal photon counter-propagate relative to the pump beam, are observed, with narrow-band peaks corresponding to 10th to 44th order quasi-phase-matching. △ Less

Submitted 4 July, 2024; originally announced July 2024.

Comments: 16 pages, 7 figures, 4 tables

The Influence of Experimental Imperfections on Photonic GHZ State Generation

Authors: Fabian Wiesner, Helen M. Chrzanowski, Gregor Pieplow, Tim Schröder, Anna Pappa, Janik Wolters

Abstract: While the advantages of photonic quantum computing, including direct compatibility with communication, are apparent, several imperfections such as loss and distinguishability presently limit actual implementations. These imperfections are unlikely to be completely eliminated, and it is therefore beneficial to investigate which of these are the most dominant and what is achievable under their prese… ▽ More While the advantages of photonic quantum computing, including direct compatibility with communication, are apparent, several imperfections such as loss and distinguishability presently limit actual implementations. These imperfections are unlikely to be completely eliminated, and it is therefore beneficial to investigate which of these are the most dominant and what is achievable under their presence. In this work, we provide an in-depth investigation of the influence of photon loss, multi-photon terms and photon distinguishability on the generation of photonic 3-partite GHZ states via established fusion protocols. We simulate the generation process for SPDC and solid-state-based single-photon sources using realistic parameters and show that different types of imperfections are dominant with respect to the fidelity and generation success probability. Our results indicate what are the dominant imperfections for the different photon sources and in which parameter regimes we can hope to implement photonic quantum computing in the near future. △ Less

Submitted 26 June, 2024; originally announced June 2024.

Comments: 18 pages, 6 figures, 3 tables

Quantum light microscopy

Authors: W. P. Bowen, Helen M. Chrzanowski, Dan Oron, Sven Ramelow, Dmitry Tabakaev, Alex Terrasson, Rob Thew

Abstract: Much of our progress in understanding microscale biology has been powered by advances in microscopy. For instance, super-resolution microscopes allow the observation of biological structures at near-atomic-scale resolution, while multi-photon microscopes allow imaging deep into tissue. However, biological structures and dynamics still often remain out of reach of existing microscopes, with further… ▽ More Much of our progress in understanding microscale biology has been powered by advances in microscopy. For instance, super-resolution microscopes allow the observation of biological structures at near-atomic-scale resolution, while multi-photon microscopes allow imaging deep into tissue. However, biological structures and dynamics still often remain out of reach of existing microscopes, with further advances in signal-to-noise, resolution and speed needed to access them. In many cases, the performance of microscopes is now limited by quantum effects -- such as noise due to the quantisation of light into photons or, for multi-photon microscopes, the low cross-section of multi-photon scattering. These limitations can be overcome by exploiting features of quantum mechanics such as entanglement. Quantum effects can also provide new ways to enhance the performance of microscopes, such as new super-resolution techniques and new techniques to image at difficult to reach wavelengths. This review provides an overview of these various ways in which quantum techniques can improve microscopy, including recent experimental progress. It seeks to provide a realistic picture of what is possible, and what the constraints and opportunities are. △ Less

Submitted 23 November, 2023; v1 submitted 9 November, 2023; originally announced November 2023.

Low-noise quantum frequency conversion in a monolithic cavity with bulk periodically poled potassium titanyl phosphate

Authors: Felix Mann, Helen M. Chrzanowski, Felipe Gewers, Marlon Placke, Sven Ramelow

Abstract: Interfacing the different building blocks of a future large scale quantum network will demand efficient and noiseless frequency conversion of quantum light. Nitrogen-vacancy (NV) centers in diamond are a leading candidate to form the nodes of such a network. However, the performance of a suitable converter remains a bottleneck, with existing demonstrations severely limited by parasitic noise arisi… ▽ More Interfacing the different building blocks of a future large scale quantum network will demand efficient and noiseless frequency conversion of quantum light. Nitrogen-vacancy (NV) centers in diamond are a leading candidate to form the nodes of such a network. However, the performance of a suitable converter remains a bottleneck, with existing demonstrations severely limited by parasitic noise arising at the target telecom wavelength. Here, we demonstrate a new platform for efficient low-noise quantum frequency conversion based on a monolithic bulk ppKTP cavity and show its suitability for the conversion of 637 nm single photons from NV centers in diamond to telecommunication wavelengths. By resonantly enhancing the power of an off-the-shelf pump laser, we achieve an internal conversion efficiency of $(72.3\pm 0.4)\%$ while generating only $(110\pm 4) \mbox{ kHz/nm}$ noise at the target wavelength without the need for any active stabilization. This constitutes a 5-fold improvement in noise over existing state-of-the-art single-step converters at this wavelengths. We verify the almost ideal preservation of non-classical correlations by converting photons from a spontaneous parametric down-conversion source and moreover show the preservation of time-energy entanglement via Franson interferometry. △ Less

Submitted 18 October, 2023; v1 submitted 26 April, 2023; originally announced April 2023.

Comments: 7 pages, 7 figures, 2 tables

Finding the Right Recipe for Low Resource Domain Adaptation in Neural Machine Translation

Authors: Virginia Adams, Sandeep Subramanian, Mike Chrzanowski, Oleksii Hrinchuk, Oleksii Kuchaiev

Abstract: General translation models often still struggle to generate accurate translations in specialized domains. To guide machine translation practitioners and characterize the effectiveness of domain adaptation methods under different data availability scenarios, we conduct an in-depth empirical exploration of monolingual and parallel data approaches to domain adaptation of pre-trained, third-party, NMT… ▽ More General translation models often still struggle to generate accurate translations in specialized domains. To guide machine translation practitioners and characterize the effectiveness of domain adaptation methods under different data availability scenarios, we conduct an in-depth empirical exploration of monolingual and parallel data approaches to domain adaptation of pre-trained, third-party, NMT models in settings where architecture change is impractical. We compare data centric adaptation methods in isolation and combination. We study method effectiveness in very low resource (8k parallel examples) and moderately low resource (46k parallel examples) conditions and propose an ensemble approach to alleviate reductions in original domain translation quality. Our work includes three domains: consumer electronic, clinical, and biomedical and spans four language pairs - Zh-En, Ja-En, Es-En, and Ru-En. We also make concrete recommendations for achieving high in-domain performance and release our consumer electronic and medical domain datasets for all languages and make our code publicly available. △ Less

Submitted 2 June, 2022; originally announced June 2022.

Mid-IR spectroscopy with NIR grating spectrometers

Authors: Paul Kaufmann, Helen M. Chrzanowski, Aron Vanselow, Sven Ramelow

Abstract: Mid-infrared (mid-IR) spectroscopy is a crucial workhorse for a plethora of analytical applications and is suitable for diverse materials, including gases, polymers or biological tissue. However, this technologically significant wavelength regime between 2.5-10$μ$m suffers from technical limitations primarily related to the large noise in mid-IR detectors and the complexity and cost of bright, bro… ▽ More Mid-infrared (mid-IR) spectroscopy is a crucial workhorse for a plethora of analytical applications and is suitable for diverse materials, including gases, polymers or biological tissue. However, this technologically significant wavelength regime between 2.5-10$μ$m suffers from technical limitations primarily related to the large noise in mid-IR detectors and the complexity and cost of bright, broadband mid-IR light sources. Here, using highly non-degenerate, broadband photon pairs from bright spontaneous parametric down-conversion (SPDC) in a nonlinear interferometer, we circumvent these limitations and realise spectroscopy in the mid-IR using only a visible (VIS) solid-state laser and an off-the-shelf, commercial near-infrared (NIR) grating spectrometer. With this proof-of-concept implementation, covering a broad range from 3.2$μ$m to 4.4$μ$m, we access short integration times down to 1s and signal-to-noise ratios above 200 at a spectral resolution from 12cm$^{-1}$ down to 1.5cm$^{-1}$ for longer integration times. Through the analysis of polymer samples and the ambient CO$_2$ in our laboratory, we highlight the potential of this measurement technique for real-world applications. △ Less

Submitted 23 September, 2021; originally announced September 2021.

Comments: 12 pages, 4 figures, submitted to Optics Express

Reconstruction of photon number conditioned states using phase randomized homodyne measurements

Authors: H. M. Chrzanowski, S. M. Assad, Julien Bernu, Boris Hage, A. P. Lund, T. C. Ralph, P. K. Lam, T. Symul

Abstract: We experimentally demonstrate the reconstruction of a photon number conditioned state without using a photon number discriminating detector. By using only phase randomized homodyne measurements, we reconstruct up to the three photon subtracted squeezed vacuum state. The reconstructed Wigner functions of these states show regions of pronounced negativity, signifying the non-classical nature of the… ▽ More We experimentally demonstrate the reconstruction of a photon number conditioned state without using a photon number discriminating detector. By using only phase randomized homodyne measurements, we reconstruct up to the three photon subtracted squeezed vacuum state. The reconstructed Wigner functions of these states show regions of pronounced negativity, signifying the non-classical nature of the reconstructed states. The techniques presented allow for complete characterization of the role of a conditional measurement on an ensemble of states, and might prove useful in systems where photon counting still proves technically challenging. △ Less

Submitted 27 July, 2021; v1 submitted 26 July, 2021; originally announced July 2021.

Journal ref: Journal of Physics B: Atomic, Molecular and Optical Physics 46, no. 10 (2013): 104009

Mid-Infrared Microscopy via Position Correlations of Undetected Photons

Authors: Inna Kviatkovsky, Helen M. Chrzanowski, Sven Ramelow

Abstract: Quantum imaging with undetected photons (QIUP) has recently emerged as a new powerful imaging tool. Exploiting the spatial entanglement of photon pairs, it allows decoupling of the sensing and detection wavelengths, facilitating imaging in otherwise challenging spectral regions with mature silicon-based detection technology. All existing implementations of QIUP have so far utilised the momentum co… ▽ More Quantum imaging with undetected photons (QIUP) has recently emerged as a new powerful imaging tool. Exploiting the spatial entanglement of photon pairs, it allows decoupling of the sensing and detection wavelengths, facilitating imaging in otherwise challenging spectral regions with mature silicon-based detection technology. All existing implementations of QIUP have so far utilised the momentum correlations within the biphoton state. Here, for the first time, we implement and examine theoretically and numerically the complementary scenario - utilising the tight position correlations formed within photon pair at birth. This image plane arrangement facilitates high resolution imaging with comparative experimental ease, and we experimentally show resolutions below 10 $μ$m at a sensing wavelength of 3.7 $μ$m. Moreover, imaging a slice of mouse heart tissue at the mid-IR to reveal morphological features on the cellular level, we further demonstrate the viability of the technique for the life sciences. These results offer new perspectives on the capabilities of QIUP for label-free wide-field microscopy, enabling new real-world applications in biomedical as well as industrial imaging at inaccessible wavelengths. △ Less

Submitted 11 June, 2021; originally announced June 2021.

Low random duty-cycle errors in periodically-poled KTP revealed by sum-frequency generation

Authors: Felix Mann, Helen M. Chrzanowski, Sven Ramelow

Abstract: Low-noise quantum frequency conversion in periodically-poled nonlinear crystals has proved challenging when the pump wavelength is shorter than the target wavelength. This is - at least in large part - a consequence of the parasitic spontaneous parametric down-conversion of pump photons, whose efficiency is increased by fabrication errors in the periodic poling. Here we characterise the poling qua… ▽ More Low-noise quantum frequency conversion in periodically-poled nonlinear crystals has proved challenging when the pump wavelength is shorter than the target wavelength. This is - at least in large part - a consequence of the parasitic spontaneous parametric down-conversion of pump photons, whose efficiency is increased by fabrication errors in the periodic poling. Here we characterise the poling quality of commercial periodically-poled bulk potassium titanyl phosphate (ppKTP) by measuring the sum-frequency generation (SFG) efficiency over a large phase mismatch range from 0 to more than 400$π$. Over the probed range the SFG efficiency behaves nearly ideally and drops to a normalised efficiency of $10^{-6}$. Our results demonstrate that any background pedestal which would be formed by random duty cycle errors in ppKTP is substantially reduced when compared to periodically poled lithium niobate. The standard deviation of the random duty cycle errors can be estimated to be smaller than 2% of the domain length. From this, we expect a noise spectral density which is at least one order of magnitude smaller than that of current state-of-the-art single-step frequency converters. △ Less

Submitted 3 June, 2021; v1 submitted 9 April, 2021; originally announced April 2021.

Comments: 5 pages, 4 figures, 1 table

Non-Attentive Tacotron: Robust and Controllable Neural TTS Synthesis Including Unsupervised Duration Modeling

Authors: Jonathan Shen, Ye Jia, Mike Chrzanowski, Yu Zhang, Isaac Elias, Heiga Zen, Yonghui Wu

Abstract: This paper presents Non-Attentive Tacotron based on the Tacotron 2 text-to-speech model, replacing the attention mechanism with an explicit duration predictor. This improves robustness significantly as measured by unaligned duration ratio and word deletion rate, two metrics introduced in this paper for large-scale robustness evaluation using a pre-trained speech recognition model. With the use of… ▽ More This paper presents Non-Attentive Tacotron based on the Tacotron 2 text-to-speech model, replacing the attention mechanism with an explicit duration predictor. This improves robustness significantly as measured by unaligned duration ratio and word deletion rate, two metrics introduced in this paper for large-scale robustness evaluation using a pre-trained speech recognition model. With the use of Gaussian upsampling, Non-Attentive Tacotron achieves a 5-scale mean opinion score for naturalness of 4.41, slightly outperforming Tacotron 2. The duration predictor enables both utterance-wide and per-phoneme control of duration at inference time. When accurate target durations are scarce or unavailable in the training data, we propose a method using a fine-grained variational auto-encoder to train the duration predictor in a semi-supervised or unsupervised manner, with results almost as good as supervised training. △ Less

Submitted 11 May, 2021; v1 submitted 8 October, 2020; originally announced October 2020.

Frequency-domain optical coherence tomography with undetected mid-infrared photons

Authors: Aron Vanselow, Paul Kaufmann, Ivan Zorin, Bettina Heise, Helen M. Chrzanowski, Sven Ramelow

Abstract: Mid-infrared light scatters much less than shorter wavelengths, allowing greatly enhanced penetration depths for optical imaging techniques such as optical coherence tomography (OCT). However, both detection and broadband sources in the mid-IR are technologically challenging. Interfering entangled photons in a nonlinear interferometer enables sensing with undetected photons making mid-IR sources a… ▽ More Mid-infrared light scatters much less than shorter wavelengths, allowing greatly enhanced penetration depths for optical imaging techniques such as optical coherence tomography (OCT). However, both detection and broadband sources in the mid-IR are technologically challenging. Interfering entangled photons in a nonlinear interferometer enables sensing with undetected photons making mid-IR sources and detectors obsolete. Here we implement mid-infrared frequency-domain OCT based on ultra-broadband entangled photon pairs. We demonstrate 10 $μ$m axial and 20 $μ$m lateral resolution 2D and 3D imaging of strongly scattering ceramic and paint samples. Together with $10^6$ times less noise scaled for the same amount of probe light and also vastly reduced footprint and technical complexity this technique can outperform conventional approaches with classical mid-IR light. △ Less

Submitted 12 June, 2020; originally announced June 2020.

Quantum-enhanced stimulated emission microscopy

Authors: Gil Triginer Garces, Helen M. Chrzanowski, Shakib Daryanoosh, Valerian Thiel, Anna L. Marchant, Raj B. Patel, Peter C. Humphreys, Animesh Datta, Ian A. Walmsley

Abstract: Nonlinear optical microscopy techniques have emerged as a set of successful tools for biological imaging. Stimulated emission microscopy belongs to a small subset of pump-probe techniques which can image non-fluorescent samples without requiring fluorescent labelling. However, its sensitivity has been shown to be ultimately limited by the quantum fluctuations in the probe beam. We propose and expe… ▽ More Nonlinear optical microscopy techniques have emerged as a set of successful tools for biological imaging. Stimulated emission microscopy belongs to a small subset of pump-probe techniques which can image non-fluorescent samples without requiring fluorescent labelling. However, its sensitivity has been shown to be ultimately limited by the quantum fluctuations in the probe beam. We propose and experimentally implement sub-shot-noise limited stimulated emission microscopy by preparing the probe pulse in an intensity-squeezed state. This technique paves the way for imaging delicate biological samples that have no detectable fluorescence with sensitivity beyond standard quantum fluctuations. △ Less

Submitted 22 April, 2020; originally announced April 2020.

Comments: The following article has been submitted to Applied Physics Letters

Journal ref: Appl. Phys. Lett. 117, 024002 (2020)

Microscopy with undetected photons in the mid-infrared

Authors: Inna Kviatkovsky, Helen M. Chrzanowski, Ellen G. Avery, Hendrik Bartolomaeus, Sven Ramelow

Abstract: Owing to its capacity for unique (bio)-chemical specificity, microscopy withmid-IR illumination holds tremendous promise for a wide range of biomedical and industrial applications. The primary limitation, however, remains detection; with current mid-IR detection technology often marrying inferior technical capabilities with prohibitive costs. This has lead to approaches that shift detection towave… ▽ More Owing to its capacity for unique (bio)-chemical specificity, microscopy withmid-IR illumination holds tremendous promise for a wide range of biomedical and industrial applications. The primary limitation, however, remains detection; with current mid-IR detection technology often marrying inferior technical capabilities with prohibitive costs. This has lead to approaches that shift detection towavelengths into the visible regime, where vastly superior silicon-based cameratechnology is available. Here, we experimentally show how nonlinear interferometry with entangled light can provide a powerful tool for mid-IR microscopy, while only requiring near-infrared detection with a standard CMOS camera. In this proof-of-principle implementation, we demonstrate intensity imaging overa broad wavelength range covering 3.4-4.3um and demonstrate a spatial resolution of 35um for images containing 650 resolved elements. Moreover, we demonstrate our technique is fit for purpose, acquiring microscopic images of biological tissue samples in the mid-IR. These results open a new perspective for potential relevance of quantum imaging techniques in the life sciences. △ Less

Submitted 29 May, 2020; v1 submitted 14 February, 2020; originally announced February 2020.

Comments: back-to-back submission with arXiv:2002.05956, Anna V. Paterova, Sivakumar M. Maniam, Hongzhi Yang, Gianluca Grenci, and Leonid A. Krivitsky, "Hyperspectral Infrared Microscopy With Visible Light "

Interfering distinguishable photons

Authors: Alex E. Jones, Adrian J. Menssen, Helen M. Chrzanowski, Tom A. W. Wolterink, Valery S. Shchesnovich, Ian A. Walmsley

Abstract: One of the central principles of quantum mechanics is that if there are multiple paths that lead to the same event, and there is no way to distinguish between them, interference occurs. It is usually assumed that distinguishing information in the preparation, evolution or measurement of a system is sufficient to destroy interference. For example, determining which slit a particle takes in Young's… ▽ More One of the central principles of quantum mechanics is that if there are multiple paths that lead to the same event, and there is no way to distinguish between them, interference occurs. It is usually assumed that distinguishing information in the preparation, evolution or measurement of a system is sufficient to destroy interference. For example, determining which slit a particle takes in Young's double slit experiment or using distinguishable photons in the two-photon Hong-Ou-Mandel effect allow discrimination of the paths leading to detection events, so in both cases interference vanishes. Remarkably for more than three independent quantum particles, distinguishability of the prepared states is not a sufficient condition for multiparticle interference to disappear. Here we experimentally demonstrate this for four photons prepared in pairwise distinguishable states, thus fundamentally challenging intuition of multiparticle interference. △ Less

Submitted 22 January, 2020; originally announced January 2020.

Comments: 23 pages, 18 figures

Journal ref: Phys. Rev. Lett. 125, 123603 (2020)

Towards Robust Image Classification Using Sequential Attention Models

Authors: Daniel Zoran, Mike Chrzanowski, Po-Sen Huang, Sven Gowal, Alex Mott, Pushmeet Kohl

Abstract: In this paper we propose to augment a modern neural-network architecture with an attention model inspired by human perception. Specifically, we adversarially train and analyze a neural model incorporating a human inspired, visual attention component that is guided by a recurrent top-down sequential process. Our experimental evaluation uncovers several notable findings about the robustness and beha… ▽ More In this paper we propose to augment a modern neural-network architecture with an attention model inspired by human perception. Specifically, we adversarially train and analyze a neural model incorporating a human inspired, visual attention component that is guided by a recurrent top-down sequential process. Our experimental evaluation uncovers several notable findings about the robustness and behavior of this new model. First, introducing attention to the model significantly improves adversarial robustness resulting in state-of-the-art ImageNet accuracies under a wide range of random targeted attack strengths. Second, we show that by varying the number of attention steps (glances/fixations) for which the model is unrolled, we are able to make its defense capabilities stronger, even in light of stronger attacks --- resulting in a "computational race" between the attacker and the defender. Finally, we show that some of the adversarial examples generated by attacking our model are quite different from conventional adversarial examples --- they contain global, salient and spatially coherent structures coming from the target class that would be recognizable even to a human, and work by distracting the attention of the model away from the main object in the original image. △ Less

Submitted 4 December, 2019; originally announced December 2019.

Ultra-broadband SPDC for spectrally far separated photon pairs

Authors: Aron Vanselow, Paul Kaufmann, Helen M. Chrzanowski, Sven Ramelow

Abstract: Generating photon pairs via spontaneous parametric down-conversion (SPDC) in nonlinear crystals is important for a wide range of quantum optics experiments with spectral properties such as their bandwidths often being a crucial concern. Here, we show the generic existence of particular phase-matching conditions in quasi-phase matched KTP, MgO:LN and SLT crystals that lead to ultra-broadband, widel… ▽ More Generating photon pairs via spontaneous parametric down-conversion (SPDC) in nonlinear crystals is important for a wide range of quantum optics experiments with spectral properties such as their bandwidths often being a crucial concern. Here, we show the generic existence of particular phase-matching conditions in quasi-phase matched KTP, MgO:LN and SLT crystals that lead to ultra-broadband, widely non-degenerate photon pairs. It is based on the existence of group-velocity matched, far apart wavelength pairs and for 2 mm long crystals results in SPDC bandwidths between 15 and 25 THz (FWHM) for photon pairs with the idler photon in the technologically relevant mid-IR range 3-5 μm and the signal photon in the NIR below 1100 nm. We experimentally demonstrate this type of broadband phase-matching in ppKTP crystals for photon pairs centered at 800 nm and 3800 nm and measure a bandwidth of 15 THz. This novel method of generating broadband photon-pairs will be highly beneficial for SPDC-based imaging, spectroscopy, refractometry and OCT with undetected mid-IR photons. △ Less

Submitted 1 August, 2019; v1 submitted 12 July, 2019; originally announced July 2019.

Comments: References added

Journal ref: Optics Letters Vol. 44, Issue 19, pp. 4638-4641 (2019)

Towards Interpretable Reinforcement Learning Using Attention Augmented Agents

Authors: Alex Mott, Daniel Zoran, Mike Chrzanowski, Daan Wierstra, Danilo J. Rezende

Abstract: Inspired by recent work in attention models for image captioning and question answering, we present a soft attention model for the reinforcement learning domain. This model uses a soft, top-down attention mechanism to create a bottleneck in the agent, forcing it to focus on task-relevant information by sequentially querying its view of the environment. The output of the attention mechanism allows… ▽ More Inspired by recent work in attention models for image captioning and question answering, we present a soft attention model for the reinforcement learning domain. This model uses a soft, top-down attention mechanism to create a bottleneck in the agent, forcing it to focus on task-relevant information by sequentially querying its view of the environment. The output of the attention mechanism allows direct observation of the information used by the agent to select its actions, enabling easier interpretation of this model than of traditional models. We analyze different strategies that the agents learn and show that a handful of strategies arise repeatedly across different games. We also show that the model learns to query separately about space and content (`where' vs. `what'). We demonstrate that an agent using this mechanism can achieve performance competitive with state-of-the-art models on ATARI tasks while still being interpretable. △ Less

Submitted 6 June, 2019; originally announced June 2019.

Learning and Evaluating General Linguistic Intelligence

Authors: Dani Yogatama, Cyprien de Masson d'Autume, Jerome Connor, Tomas Kocisky, Mike Chrzanowski, Lingpeng Kong, Angeliki Lazaridou, Wang Ling, Lei Yu, Chris Dyer, Phil Blunsom

Abstract: We define general linguistic intelligence as the ability to reuse previously acquired knowledge about a language's lexicon, syntax, semantics, and pragmatic conventions to adapt to new tasks quickly. Using this definition, we analyze state-of-the-art natural language understanding models and conduct an extensive empirical investigation to evaluate them against these criteria through a series of ex… ▽ More We define general linguistic intelligence as the ability to reuse previously acquired knowledge about a language's lexicon, syntax, semantics, and pragmatic conventions to adapt to new tasks quickly. Using this definition, we analyze state-of-the-art natural language understanding models and conduct an extensive empirical investigation to evaluate them against these criteria through a series of experiments that assess the task-independence of the knowledge being acquired by the learning process. In addition to task performance, we propose a new evaluation metric based on an online encoding of the test data that quantifies how quickly an existing agent (model) learns a new task. Our results show that while the field has made impressive progress in terms of model architectures that generalize to many tasks, these models still require a lot of in-domain training examples (e.g., for fine tuning, training task-specific modules), and are prone to catastrophic forgetting. Moreover, we find that far from solving general tasks (e.g., document question answering), our models are overfitting to the quirks of particular datasets (e.g., SQuAD). We discuss missing components and conjecture on how to make progress toward general linguistic intelligence. △ Less

Submitted 31 January, 2019; originally announced January 2019.

Entanglement properties of a measurement-based entanglement distillation experiment

Authors: Hao Jeng, Spyros Tserkis, **g Yan Haw, Helen M. Chrzanowski, Jiri Janousek, Timothy C. Ralph, ** Koy Lam, Syed M. Assad

Abstract: Measures of entanglement can be employed for the analysis of numerous quantum information protocols. Due to computational convenience, logarithmic negativity is often the choice in the case of continuous variable systems. In this work, we analyse a continuous variable measurement-based entanglement distillation experiment using a collection of entanglement measures. This includes: logarithmic nega… ▽ More Measures of entanglement can be employed for the analysis of numerous quantum information protocols. Due to computational convenience, logarithmic negativity is often the choice in the case of continuous variable systems. In this work, we analyse a continuous variable measurement-based entanglement distillation experiment using a collection of entanglement measures. This includes: logarithmic negativity, entanglement of formation, distillable entanglement, relative entropy of entanglement, and squashed entanglement. By considering the distilled entanglement as a function of the success probability of the distillation protocol, we show that the logarithmic negativity surpasses the bound on deterministic entanglement distribution at a relatively large probability of success. This is in contrast to the other measures which would only be able to do so at much lower probabilities, hence demonstrating that logarithmic negativity alone is inadequate for assessing the performance of the distillation protocol. In addition to this result, we also observed an increase in the distillable entanglement by making use of upper and lower bounds to estimate this quantity. We thus demonstrate the utility of these theoretical tools in an experimental setting. △ Less

Submitted 27 November, 2018; originally announced November 2018.

Comments: 12 pages, 8 figures, submitted

Journal ref: Phys. Rev. A 99, 042304 (2019)

Relational recurrent neural networks

Authors: Adam Santoro, Ryan Faulkner, David Raposo, Jack Rae, Mike Chrzanowski, Theophane Weber, Daan Wierstra, Oriol Vinyals, Razvan Pascanu, Timothy Lillicrap

Abstract: Memory-based neural networks model temporal data by leveraging an ability to remember information for long periods. It is unclear, however, whether they also have an ability to perform complex relational reasoning with the information they remember. Here, we first confirm our intuitions that standard memory architectures may struggle at tasks that heavily involve an understanding of the ways in wh… ▽ More Memory-based neural networks model temporal data by leveraging an ability to remember information for long periods. It is unclear, however, whether they also have an ability to perform complex relational reasoning with the information they remember. Here, we first confirm our intuitions that standard memory architectures may struggle at tasks that heavily involve an understanding of the ways in which entities are connected -- i.e., tasks involving relational reasoning. We then improve upon these deficits by using a new memory module -- a \textit{Relational Memory Core} (RMC) -- which employs multi-head dot product attention to allow memories to interact. Finally, we test the RMC on a suite of tasks that may profit from more capable relational reasoning across sequential information, and show large gains in RL domains (e.g. Mini PacMan), program evaluation, and language modeling, achieving state-of-the-art results on the WikiText-103, Project Gutenberg, and GigaWord datasets. △ Less

Submitted 28 June, 2018; v1 submitted 5 June, 2018; originally announced June 2018.

Unconditional violation of the shot noise limit in photonic quantum metrology

Authors: Sergei Slussarenko, Morgan M. Weston, Helen M. Chrzanowski, Lynden K. Shalm, Varun B. Verma, Sae Woo Nam, Geoff J. Pryde

Abstract: Interferometric phase measurement is widely used to precisely determine quantities such as length, speed, and material properties. Without quantum correlations, the best phase sensitivity $Δ\varphi$ achievable using $n$ photons is the shot noise limit (SNL), $Δ\varphi=1/\sqrt{n}$. Quantum-enhanced metrology promises better sensitivity, but despite theoretical proposals stretching back decades, no… ▽ More Interferometric phase measurement is widely used to precisely determine quantities such as length, speed, and material properties. Without quantum correlations, the best phase sensitivity $Δ\varphi$ achievable using $n$ photons is the shot noise limit (SNL), $Δ\varphi=1/\sqrt{n}$. Quantum-enhanced metrology promises better sensitivity, but despite theoretical proposals stretching back decades, no measurement using photonic (i.e. definite photon number) quantum states has truly surpassed the SNL. Rather, all such demonstrations --- by discounting photon loss, detector inefficiency, or other imperfections --- have considered only a subset of the photons used. Here, we use an ultra-high efficiency photon source and detectors to perform unconditional entanglement-enhanced photonic interferometry. Sampling a birefringent phase shift, we demonstrate precision beyond the SNL without artificially correcting our results for loss and imperfections. Our results enable quantum-enhanced phase measurements at low photon flux and open the door to the next generation of optical quantum metrology advances. △ Less

Submitted 22 June, 2018; v1 submitted 27 July, 2017; originally announced July 2017.

Comments: 5 pages,3 figures

Journal ref: Nature Photonics 11, 700-703 (2017)

Gaussian Optical Ising Machines

Authors: William R. Clements, Jelmer J. Renema, Y. Henry Wen, Helen M. Chrzanowski, W. Steven Kolthammer, Ian A. Walmsley

Abstract: It has recently been shown that optical parametric oscillator (OPO) Ising machines, consisting of coupled optical pulses circulating in a cavity with parametric gain, can be used to probabilistically find low-energy states of Ising spin systems. In this work, we study optical Ising machines that operate under simplified Gaussian dynamics. We show that these dynamics are sufficient for reaching pro… ▽ More It has recently been shown that optical parametric oscillator (OPO) Ising machines, consisting of coupled optical pulses circulating in a cavity with parametric gain, can be used to probabilistically find low-energy states of Ising spin systems. In this work, we study optical Ising machines that operate under simplified Gaussian dynamics. We show that these dynamics are sufficient for reaching probabilities of success comparable to previous work. Based on this result, we propose modified optical Ising machines with simpler designs that do not use parametric gain yet achieve similar performance, thus suggesting a route to building much larger systems. △ Less

Submitted 20 June, 2017; originally announced June 2017.

Comments: 6 pages

Journal ref: Phys. Rev. A 96, 043850 (2017)

Deep Voice: Real-time Neural Text-to-Speech

Authors: Sercan O. Arik, Mike Chrzanowski, Adam Coates, Gregory Diamos, Andrew Gibiansky, Yongguo Kang, Xian Li, John Miller, Andrew Ng, Jonathan Raiman, Shubho Sengupta, Mohammad Shoeybi

Abstract: We present Deep Voice, a production-quality text-to-speech system constructed entirely from deep neural networks. Deep Voice lays the groundwork for truly end-to-end neural speech synthesis. The system comprises five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency predi… ▽ More We present Deep Voice, a production-quality text-to-speech system constructed entirely from deep neural networks. Deep Voice lays the groundwork for truly end-to-end neural speech synthesis. The system comprises five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. For the segmentation model, we propose a novel way of performing phoneme boundary detection with deep neural networks using connectionist temporal classification (CTC) loss. For the audio synthesis model, we implement a variant of WaveNet that requires fewer parameters and trains faster than the original. By using a neural network for each component, our system is simpler and more flexible than traditional text-to-speech systems, where each component requires laborious feature engineering and extensive domain expertise. Finally, we show that inference with our system can be performed faster than real time and describe optimized WaveNet inference kernels on both CPU and GPU that achieve up to 400x speedups over existing implementations. △ Less

Submitted 7 March, 2017; v1 submitted 24 February, 2017; originally announced February 2017.

Comments: Submitted to ICML 2017

Heralded quantum steering over a high-loss channel

Authors: Morgan M. Weston, Sergei Slussarenko, Helen M. Chrzanowski, Sabine Wollmann, Lynden K. Shalm, Varun B. Verma, Michael S. Allman, Sae Woo Nam, Geoff J. Pryde

Abstract: Entanglement is the key resource for many long-range quantum information tasks, including secure communication and fundamental tests of quantum physics. These tasks require robust verification of shared entanglement, but performing it over long distances is presently technologically intractable because the loss through an optical fiber or free-space channel opens up a detection loophole. We design… ▽ More Entanglement is the key resource for many long-range quantum information tasks, including secure communication and fundamental tests of quantum physics. These tasks require robust verification of shared entanglement, but performing it over long distances is presently technologically intractable because the loss through an optical fiber or free-space channel opens up a detection loophole. We design and experimentally demonstrate a scheme that verifies entanglement in the presence of at least $14.8\pm0.1$ dB of added loss, equivalent to approximately $80$ km of telecommunication fiber. Our protocol relies on entanglement swap** to herald the presence of a photon after the lossy channel, enabling event-ready implementation of quantum steering. This result overcomes the key barrier in device-independent communication under realistic high-loss scenarios and in the realization of a quantum repeater. △ Less

Submitted 9 January, 2018; v1 submitted 20 December, 2016; originally announced December 2016.

Comments: 8 pages, 5 figures

Journal ref: Sci. Adv. 4, e1701230 (2018)

Efficient and pure femtosecond-pulse-length source of polarization-entangled photons

Authors: Morgan M. Weston, Helen M. Chrzanowski, Sabine Wollmann, Allen Boston, Joseph Ho, Lynden K. Shalm, Varun B. Verma, Michael S. Allman, Sae Woo Nam, Raj B. Patel, Sergei Slussarenko, Geoff J. Pryde

Abstract: We present a source of polarization entangled photon pairs based on spontaneous parametric downconversion engineered for frequency uncorrelated telecom photon generation. Our source provides photon pairs that display, simultaneously, the key properties for high-performance quantum information and fundamental quantum science tasks. Specifically, the source provides for high heralding efficiency, hi… ▽ More We present a source of polarization entangled photon pairs based on spontaneous parametric downconversion engineered for frequency uncorrelated telecom photon generation. Our source provides photon pairs that display, simultaneously, the key properties for high-performance quantum information and fundamental quantum science tasks. Specifically, the source provides for high heralding efficiency, high quantum state purity and high entangled state fidelity at the same time. Among different tests we apply to our source we observe almost perfect non-classical interference between photons from independent sources with a visibility of $(100\pm5)\%$. △ Less

Submitted 11 May, 2016; v1 submitted 11 March, 2016; originally announced March 2016.

Comments: 16 pages, 4 figures

Journal ref: Opt. Express 24 (10), pp. 10869-10879 (2016)

Deep Speech 2: End-to-End Speech Recognition in English and Mandarin

Authors: Dario Amodei, Rishita Anubhai, Eric Battenberg, Carl Case, Jared Casper, Bryan Catanzaro, **gdong Chen, Mike Chrzanowski, Adam Coates, Greg Diamos, Erich Elsen, Jesse Engel, Linxi Fan, Christopher Fougner, Tony Han, Awni Hannun, Billy Jun, Patrick LeGresley, Libby Lin, Sharan Narang, Andrew Ng, Sherjil Ozair, Ryan Prenger, Jonathan Raiman, Sanjeev Satheesh , et al. (9 additional authors not shown)

Abstract: We show that an end-to-end deep learning approach can be used to recognize either English or Mandarin Chinese speech--two vastly different languages. Because it replaces entire pipelines of hand-engineered components with neural networks, end-to-end learning allows us to handle a diverse variety of speech including noisy environments, accents and different languages. Key to our approach is our app… ▽ More We show that an end-to-end deep learning approach can be used to recognize either English or Mandarin Chinese speech--two vastly different languages. Because it replaces entire pipelines of hand-engineered components with neural networks, end-to-end learning allows us to handle a diverse variety of speech including noisy environments, accents and different languages. Key to our approach is our application of HPC techniques, resulting in a 7x speedup over our previous system. Because of this efficiency, experiments that previously took weeks now run in days. This enables us to iterate more quickly to identify superior architectures and algorithms. As a result, in several cases, our system is competitive with the transcription of human workers when benchmarked on standard datasets. Finally, using a technique called Batch Dispatch with GPUs in the data center, we show that our system can be inexpensively deployed in an online setting, delivering low latency when serving users at scale. △ Less

Submitted 8 December, 2015; originally announced December 2015.

Measurement-Based Noiseless Linear Amplification for Quantum Communication

Authors: Helen M. Chrzanowski, Nathan Walk, Syed M. Assad, Jiri Janousek, Sara Hosseini, Timothy C. Ralph, Thomas Symul, ** Koy Lam

Abstract: Entanglement distillation is an indispensable ingredient in extended quantum communication networks. Distillation protocols are necessarily non-deterministic and require advanced experimental techniques such as noiseless amplification. Recently it was shown that the benefits of noiseless amplification could be extracted by performing a post-selective filtering of the measurement record to improve… ▽ More Entanglement distillation is an indispensable ingredient in extended quantum communication networks. Distillation protocols are necessarily non-deterministic and require advanced experimental techniques such as noiseless amplification. Recently it was shown that the benefits of noiseless amplification could be extracted by performing a post-selective filtering of the measurement record to improve the performance of quantum key distribution. We apply this protocol to entanglement degraded by transmission loss of up to the equivalent of 100km of optical fibre. We measure an effective entangled resource stronger than that achievable by even a maximally entangled resource passively transmitted through the same channel. We also provide a proof-of-principle demonstration of secret key extraction from an otherwise insecure regime. The measurement-based noiseless linear amplifier offers two advantages over its physical counterpart: ease of implementation and near optimal probability of success. It should provide an effective and versatile tool for a broad class of entanglement-based quantum communication protocols. △ Less

Submitted 12 September, 2014; v1 submitted 9 September, 2014; originally announced September 2014.

Comments: 7+3 pages, 5+1 figures, close to published version

Journal ref: Nature Photonics, 8, 333 (2014)

Experimental Verification of Quantum Discord in Continuous-Variable States

Authors: S. Hosseini, S. Rahimi-Keshari, J. Y. Haw, S. M. Assad, H. M. Chrzanowski, J. Janousek, T. Symul, T. C. Ralph, P. K. Lam

Abstract: We introduce a simple and efficient technique to verify quantum discord in unknown Gaussian states and a certain class of non-Gaussian states. We show that any separation in the peaks of the marginal distributions of one subsystem conditioned on two different outcomes of homodyne measurements performed on the other subsystem indicates correlation between the corresponding quadratures, and hence no… ▽ More We introduce a simple and efficient technique to verify quantum discord in unknown Gaussian states and a certain class of non-Gaussian states. We show that any separation in the peaks of the marginal distributions of one subsystem conditioned on two different outcomes of homodyne measurements performed on the other subsystem indicates correlation between the corresponding quadratures, and hence nonzero discord. We also apply this method to non-Gaussian states that are prepared by overlap** a statistical mixture of coherent and vacuum states on a beam splitter. We experimentally demonstrate this technique by verifying nonzero quantum discord in a bipartite Gaussian and certain class of non-Gaussian states. △ Less

Submitted 13 January, 2014; v1 submitted 24 October, 2013; originally announced October 2013.

Comments: 14 pages, 6 figures

Journal ref: J. Phys. B: At. Mol. Opt. Phys. 47(2014)025503

Operational Significance of Discord Consumption: Theory and Experiment

Authors: Mile Gu, Helen M. Chrzanowski, Syed M. Assad, Thomas Symul, Kavan Modi, Timothy C. Ralph, Vlatko Vedral, ** Koy Lam

Abstract: Coherent interactions that generate negligible entanglement can still exhibit unique quantum behaviour. This observation has motivated a search beyond entanglement for a complete description of all quantum correlations. Quantum discord is a promising candidate. Here, we demonstrate that under certain measurement constraints, discord between bipartite systems can be consumed to encode information t… ▽ More Coherent interactions that generate negligible entanglement can still exhibit unique quantum behaviour. This observation has motivated a search beyond entanglement for a complete description of all quantum correlations. Quantum discord is a promising candidate. Here, we demonstrate that under certain measurement constraints, discord between bipartite systems can be consumed to encode information that can only be accessed by coherent quantum interactions. The inability to access this information by any other means allows us to use discord to directly quantify this `quantum advantage'. We experimentally encode information within the discordant correlations of two separable Gaussian states. The amount of extra information recovered by coherent interaction is quantified and directly linked with the discord consumed during encoding. No entanglement exists at any point of this experiment. Thus we introduce and demonstrate an operational method to use discord as a physical resource. △ Less

Submitted 4 August, 2012; v1 submitted 29 February, 2012; originally announced March 2012.

Comments: 10 pages, 3 figures, updated with Nature Physics Reference, simplified proof in Appendix

Journal ref: Nature Physics 8, 671-675 (2012)

A Scalable, Self-Analyzing Digital Locking System for use on Quantum Optics Experiments

Authors: B. M. Sparkes, H. M. Chrzanowski, D. P. Parrain, B. C. Buchler, P. K. Lam, T. Symul

Abstract: Digital control of optics experiments has many advantages over analog control systems, specifically in terms of scalability, cost, flexibility, and the integration of system information into one location. We present a digital control system, freely available for download online, specifically designed for quantum optics experiments that allows for automatic and sequential re-locking of optical comp… ▽ More Digital control of optics experiments has many advantages over analog control systems, specifically in terms of scalability, cost, flexibility, and the integration of system information into one location. We present a digital control system, freely available for download online, specifically designed for quantum optics experiments that allows for automatic and sequential re-locking of optical components. We show how the inbuilt locking analysis tools, including a white-noise network analyzer, can be used to help optimize individual locks, and verify the long term stability of the digital system. Finally, we present an example of the benefits of digital locking for quantum optics by applying the code to a specific experiment used to characterize optical Schrodinger cat states. △ Less

Submitted 27 May, 2011; v1 submitted 19 May, 2011; originally announced May 2011.

Comments: 7 pages, 5 figures

Journal ref: Rev. Sci. Instrum. 82, 075113 (2011)

Demonstrating various quantum effects with two entangled laser beams

Authors: B. Hage, J. Janoušek, S. Armstrong, T. Symul, J. Bernu, H. M. Chrzanowski, P. K. Lam, H. A. Bachor

Abstract: We report on the preparation of entangled two mode squeezed states of yet unseen quality. Based on a measurement of the covariance matrix we found a violation of the Reid and Drummond EPR-criterion at a value of only 0.36\pm0.03 compared to the threshold of 1. Furthermore, quantum state tomography was used to extract a single photon Fock state solely based on homodyne detection, demonstrating the… ▽ More We report on the preparation of entangled two mode squeezed states of yet unseen quality. Based on a measurement of the covariance matrix we found a violation of the Reid and Drummond EPR-criterion at a value of only 0.36\pm0.03 compared to the threshold of 1. Furthermore, quantum state tomography was used to extract a single photon Fock state solely based on homodyne detection, demonstrating the strong quantum features of this pair of laser-beams. The probability for a single photon in this ensemble measurement exceeded 2/3. △ Less

Submitted 2 May, 2011; v1 submitted 22 March, 2011; originally announced March 2011.

Photon number discrimination without a photon counter and its application to reconstructing non-Gaussian states

Authors: H. M. Chrzanowski, J. Bernu, B. M. Sparkes, B. Hage, A. P. Lund, T. C. Ralph, P. K. Lam, T. Symul

Abstract: The non-linearity of a conditional photon-counting measurement can be used to `de-Gaussify' a Gaussian state of light. Here we present and experimentally demonstrate a technique for photon number resolution using only homodyne detection. We then apply this technique to inform a conditional measurement; unambiguously reconstructing the statistics of the non-Gaussian one and two photon subtracted sq… ▽ More The non-linearity of a conditional photon-counting measurement can be used to `de-Gaussify' a Gaussian state of light. Here we present and experimentally demonstrate a technique for photon number resolution using only homodyne detection. We then apply this technique to inform a conditional measurement; unambiguously reconstructing the statistics of the non-Gaussian one and two photon subtracted squeezed vacuum states. Although our photon number measurement relies on ensemble averages and cannot be used to prepare non-Gaussian states of light, its high efficiency, photon number resolving capabilities, and compatibility with the telecommunications band make it suitable for quantum information tasks relying on the outcomes of mean values. △ Less

Submitted 7 December, 2011; v1 submitted 27 February, 2011; originally announced February 2011.

Comments: 4 pages, 3 figures. Theory section expanded in response to referee comments

Journal ref: Phys. Rev. A 84, 050302(R) (2011)