Search

The present paper proposes a robust evaluation of any radial density at small distances using negative-order radial moments evaluated in momentum space. This evaluation provides a valuable insight into the behavior of a given radial density in the vicinity of $r=0$, and puts strong emphasis on the importance of measuring form factors at large squared four-momentum transfer, a domain essential for the determination of negative order moments. A specific attention is paid to the regularization scheme directly affecting the numerical determination of the radial density's parametrization. The proposed method is applied to non-relativistic study cases of the nucleon electric ($G_{En}, G_{Ep}$), and proton magnetic $G_{Mp}$ form factors. The validation is performed through comparison of the results of the approach to the analytically determined Maclaurin expansion - in the vicinity of $r=0$ - of the radial density function. The method is also applied to the relativistic Dirac form factor $F_1$ of the proton. In such a non-trivial case, the Maclaurin development might not exist for the radial density, rendering the determination from the proposed method extremely important., Comment: 11 pages, 8 plots

The precision of experimental data and analysis techniques is a key feature of any discovery attempt. A striking example is the proton radius puzzle where the accuracy of the spectroscopy of muonic atoms challenges traditional electron scattering measurements. The present work proposes a novel method for the determination of spatial moments from densities expressed in the momentum space. This method provides a direct access to even, odd, and more generally any real, negative and positive moment with order larger than $-3$. As an illustration, the application of this method to the electric form factor of the proton is discussed in detail.

Voice assistants like Amazon's Alexa, Google's Assistant, or Apple's Siri, have become the primary (voice) interface in smart speakers that can be found in millions of households. For privacy reasons, these speakers analyze every sound in their environment for their respective wake word like ''Alexa'' or ''Hey Siri,'' before uploading the audio stream to the cloud for further processing. Previous work reported on the inaccurate wake word detection, which can be tricked using similar words or sounds like ''cocaine noodles'' instead of ''OK Google.'' In this paper, we perform a comprehensive analysis of such accidental triggers, i.,e., sounds that should not have triggered the voice assistant, but did. More specifically, we automate the process of finding accidental triggers and measure their prevalence across 11 smart speakers from 8 different manufacturers using everyday media such as TV shows, news, and other kinds of audio datasets. To systematically detect accidental triggers, we describe a method to artificially craft such triggers using a pronouncing dictionary and a weighted, phone-based Levenshtein distance. In total, we have found hundreds of accidental triggers. Moreover, we explore potential gender and language biases and analyze the reproducibility. Finally, we discuss the resulting privacy implications of accidental triggers and explore countermeasures to reduce and limit their impact on users' privacy. To foster additional research on these sounds that mislead machine learning models, we publish a dataset of more than 1000 verified triggers as a research artifact.

We report on a new measurement of the beam transverse single spin asymmetry in electron-proton elastic scattering, $A^{ep}_{\perp}$, at five beam energies from 315.1 MeV to 1508.4 MeV and at a scattering angle of $30^{\circ} < \theta < 40^{\circ}$. The covered $Q^2$ values are 0.032, 0.057, 0.082, 0.218, 0.613 (GeV/c)$^2$. The measurement clearly indicates significant inelastic contributions to the two-photon-exchange (TPE) amplitude in the low-$Q^2$ kinematic region. No theoretical calculation is able to reproduce our result. Comparison with a calculation based on unitarity, which only takes into account elastic and $\mathrm{\pi N}$ inelastic intermediate states, suggests that there are other inelastic intermediate states such as $\mathrm{\pi \pi N}$, $\mathrm{K \Lambda}$ and $\mathrm{\eta N}$. Covering a wide energy range, our new high-precision data provide a benchmark to study those intermediate states., Comment: 6 pages, 3 figures

Applying Q-learning to high-dimensional or continuous action spaces can be difficult due to the required maximization over the set of possible actions. Motivated by techniques from amortized inference, we replace the expensive maximization over all actions with a maximization over a small subset of possible actions sampled from a learned proposal distribution. The resulting approach, which we dub Amortized Q-learning (AQL), is able to handle discrete, continuous, or hybrid action spaces while maintaining the benefits of Q-learning. Our experiments on continuous control tasks with up to 21 dimensional actions show that AQL outperforms D3PG (Barth-Maron et al, 2018) and QT-Opt (Kalashnikov et al, 2018). Experiments on structured discrete action spaces demonstrate that AQL can efficiently learn good policies in spaces with thousands of discrete actions., Comment: A previous version of this work appeared at the Deep Reinforcement Learning Workshop, NeurIPS 2018

PANDA (anti-Proton ANnihiliation at DArmstadt) is planned to be one of the four main experiments at the future international accelerator complex FAIR (Facility for Antiproton and Ion Research) in Darmstadt, Germany. It is going to address fundamental questions of hadron physics and quantum chromodynamics using cooled antiproton beams with a high intensity and and momenta between 1.5 and 15 GeV/c. PANDA is designed to reach a maximum luminosity of 2x10^32 cm^2 s. Most of the physics programs require an excellent particle identification (PID). The PID of hadronic states at the forward endcap of the target spectrometer will be done by a fast and compact Cherenkov detector that uses the detection of internally reflected Cherenkov light (DIRC) principle. It is designed to cover the polar angle range from 5{\deg} to 22{\deg} and to provide a separation power for the separation of charged pions and kaons up to 3 standard deviations (s.d.) for particle momenta up to 4 GeV/c in order to cover the important particle phase space. This document describes the technical design and the expected performance of the novel PANDA Disc DIRC detector that has not been used in any other high energy physics experiment (HEP) before. The performance has been studied with Monte-Carlo simulations and various beam tests at DESY and CERN. The final design meets all PANDA requirements and guarantees suffcient safety margins., Comment: TDR for Panda/Fair to be published

It has been established that diverse behaviors spanning the controllable subspace of an Markov decision process can be trained by rewarding a policy for being distinguishable from other policies \citep{gregor2016variational, eysenbach2018diversity, warde2018unsupervised}. However, one limitation of this formulation is generalizing behaviors beyond the finite set being explicitly learned, as is needed for use on subsequent tasks. Successor features \citep{dayan93improving, barreto2017successor} provide an appealing solution to this generalization problem, but require defining the reward function as linear in some grounded feature space. In this paper, we show that these two techniques can be combined, and that each method solves the other's primary limitation. To do so we introduce Variational Intrinsic Successor FeatuRes (VISR), a novel algorithm which learns controllable features that can be leveraged to provide enhanced generalization and fast task inference through the successor feature framework. We empirically validate VISR on the full Atari suite, in a novel setup wherein the rewards are only exposed briefly after a long unsupervised phase. Achieving human-level performance on 14 games and beating all baselines, we believe VISR represents a step towards agents that rapidly learn from limited feedback., Comment: Accepted for publication at ICLR 2020

Learning to control an environment without hand-crafted rewards or expert data remains challenging and is at the frontier of reinforcement learning research. We present an unsupervised learning algorithm to train agents to achieve perceptually-specified goals using only a stream of observations and actions. Our agent simultaneously learns a goal-conditioned policy and a goal achievement reward function that measures how similar a state is to the goal state. This dual optimization leads to a co-operative game, giving rise to a learned reward function that reflects similarity in controllable aspects of the environment instead of distance in the space of observations. We demonstrate the efficacy of our agent to learn, in an unsupervised manner, to reach a diverse set of goals on three domains -- Atari, the DeepMind Control Suite and DeepMind Lab., Comment: 10 pages + references & 5 page appendix

The question of the determination of the proton charge radius $R_p$ from electron scattering data led to an unprecedented experimental effort for measurements of the electric form factor of the proton at low and very low momentum transfer in electron and muon elastic scattering. On the basis of basic properties of densities and fitting bias considerations, a procedure is developed in order to evaluate the impact of forthcoming data on $R_p$. Particularly, it is shown that a 0.1\% precision on these future cross section data is necessary to establish indisputably the $R_p$-value as determined from lepton scattering. The ProRad (Proton Radius) experiment at the PRAE (Platform for Research and Applications with Electrons) facility in Orsay is further discussed, especially the experimental method to meet this stringent constraint., Comment: 10 pages, 7 figures

We propose Scheduled Auxiliary Control (SAC-X), a new learning paradigm in the context of Reinforcement Learning (RL). SAC-X enables learning of complex behaviors - from scratch - in the presence of multiple sparse reward signals. To this end, the agent is equipped with a set of general auxiliary tasks, that it attempts to learn simultaneously via off-policy RL. The key idea behind our method is that active (learned) scheduling and execution of auxiliary policies allows the agent to efficiently explore its environment - enabling it to excel at sparse reward RL. Our experiments in several challenging robotic manipulation settings demonstrate the power of our approach., Comment: A video of the rich set of learned behaviours can be found at https://youtu.be/mPKyvocNe_M