-
Probing proton cross-shell excitations through the two-neutron removal from 38Ca
Authors:
T. Beck,
A. Gade,
B. A. Brown,
J. A. Tostevin,
D. Weisshaar,
D. Bazin,
K. W. Brown,
R. J. Charity,
P. J. Farris,
S. A. Gillespie,
A. M. Hill,
J. Li,
B. Longfellow,
W. Reviol,
D. Rhodes
Abstract:
Bound states of the neutron-deficient, near-dripline nucleus $^{36}$Ca were populated in two-neutron removal from the ground state of $^{38}$Ca, a direct reaction sensitive to the single-particle configurations and couplings of the removed neutrons in the projectile wave function. Final-state exclusive cross sections for the formation of $^{36}$Ca and the corresponding longitudinal momentum distri…
▽ More
Bound states of the neutron-deficient, near-dripline nucleus $^{36}$Ca were populated in two-neutron removal from the ground state of $^{38}$Ca, a direct reaction sensitive to the single-particle configurations and couplings of the removed neutrons in the projectile wave function. Final-state exclusive cross sections for the formation of $^{36}$Ca and the corresponding longitudinal momentum distributions, both determined through the combination of particle and $γ$-ray spectroscopy, are compared to predictions combining eikonal reaction theory and shell-model two-nucleon amplitudes from the USDB, USDC, and ZBM2 effective interactions. The final-state cross-section ratio $σ(2^+_1)/σ(0^+)$ shows particular sensitivity and is approximately reproduced only with the two-nucleon amplitudes from the ZBM2 effective interaction that includes proton cross-shell excitations into the $pf$ shell. Characterizing the proton $pf$-shell occupancy locally and schematically, an increase of the $sd$-$pf$ shell gap by $250$ keV yields an improved description of this cross-section ratio and simultaneously enables a reproduction of the $B(E2;0^+_1\to2^+_1)$ excitation strength of $^{36}$Ca. This highlights an important aspect if a new shell-model effective interaction for the region was to be developed on the quest to model the neutron-deficient Ca isotopes and surrounding nuclei whose structure is impacted by proton cross-shell excitations.
△ Less
Submitted 15 January, 2024;
originally announced January 2024.
-
Mask The Bias: Improving Domain-Adaptive Generalization of CTC-based ASR with Internal Language Model Estimation
Authors:
Nilaksh Das,
Monica Sunkara,
Sravan Bodapati,
**glun Cai,
Devang Kulshreshtha,
Jeff Farris,
Katrin Kirchhoff
Abstract:
End-to-end ASR models trained on large amount of data tend to be implicitly biased towards language semantics of the training data. Internal language model estimation (ILME) has been proposed to mitigate this bias for autoregressive models such as attention-based encoder-decoder and RNN-T. Typically, ILME is performed by modularizing the acoustic and language components of the model architecture,…
▽ More
End-to-end ASR models trained on large amount of data tend to be implicitly biased towards language semantics of the training data. Internal language model estimation (ILME) has been proposed to mitigate this bias for autoregressive models such as attention-based encoder-decoder and RNN-T. Typically, ILME is performed by modularizing the acoustic and language components of the model architecture, and eliminating the acoustic input to perform log-linear interpolation with the text-only posterior. However, for CTC-based ASR, it is not as straightforward to decouple the model into such acoustic and language components, as CTC log-posteriors are computed in a non-autoregressive manner. In this work, we propose a novel ILME technique for CTC-based ASR models. Our method iteratively masks the audio timesteps to estimate a pseudo log-likelihood of the internal LM by accumulating log-posteriors for only the masked timesteps. Extensive evaluation across multiple out-of-domain datasets reveals that the proposed approach improves WER by up to 9.8% and OOV F1-score by up to 24.6% relative to Shallow Fusion, when only text data from target domain is available. In the case of zero-shot domain adaptation, with no access to any target domain data, we demonstrate that removing the source domain bias with ILME can still outperform Shallow Fusion to improve WER by up to 9.3% relative.
△ Less
Submitted 5 May, 2023;
originally announced May 2023.
-
Hexadecapole strength in the rare isotopes $^{74,76}$Kr
Authors:
M. Spieker,
S. E. Agbemava,
D. Bazin,
S. Biswas P. D. Cottle,
P. J. Farris,
A. Gade,
T. Ginter,
S. Giraud,
K. W. Kemper,
J. Li,
W. Nazarewicz,
S. Noji,
J. Pereira,
L. A. Riley,
M. Smith,
D. Weisshaar,
R. G. T. Zegers
Abstract:
In the Ge-Sr mass region, isotopes with neutron number $N \leq 40$ are known to feature rapid shape changes with both nucleon number and angular momentum. To gain new insights into their structure, inelastic proton scattering experiments in inverse kinematics were performed on the rare isotopes $^{74,76}$Kr. This work focuses on observables related to the $J^π = 4^+_1$ states of the Kr isotopes an…
▽ More
In the Ge-Sr mass region, isotopes with neutron number $N \leq 40$ are known to feature rapid shape changes with both nucleon number and angular momentum. To gain new insights into their structure, inelastic proton scattering experiments in inverse kinematics were performed on the rare isotopes $^{74,76}$Kr. This work focuses on observables related to the $J^π = 4^+_1$ states of the Kr isotopes and, in particular, on the hexadecapole degree of freedom. By performing coupled-channels calculations, hexadecapole deformation parameters $β_4$ were determined for the $J^π = 4^+_1$ states of $^{74,76}$Kr from inelastic proton scattering cross sections. Two possible coupled-channels solutions were found. A comparison to predictions from nuclear energy density functional theory, employing both non-relativistic and relativistic functionals, clearly favors the large, positive $β_4$ solutions. These $β_4$ values are unambiguously linked to the well deformed prolate configuration. Given the $β_2 - β_4$ trend, established in this work, it appears that $β_4$ values could provide a sensitive measure of the nuclear shell structure.
△ Less
Submitted 27 April, 2023;
originally announced April 2023.
-
Dynamic Chunk Convolution for Unified Streaming and Non-Streaming Conformer ASR
Authors:
Xilai Li,
Goeric Huybrechts,
Srikanth Ronanki,
Jeff Farris,
Sravan Bodapati
Abstract:
Recently, there has been an increasing interest in unifying streaming and non-streaming speech recognition models to reduce development, training and deployment cost. The best-known approaches rely on either window-based or dynamic chunk-based attention strategy and causal convolutions to minimize the degradation due to streaming. However, the performance gap still remains relatively large between…
▽ More
Recently, there has been an increasing interest in unifying streaming and non-streaming speech recognition models to reduce development, training and deployment cost. The best-known approaches rely on either window-based or dynamic chunk-based attention strategy and causal convolutions to minimize the degradation due to streaming. However, the performance gap still remains relatively large between non-streaming and a full-contextual model trained independently. To address this, we propose a dynamic chunk-based convolution replacing the causal convolution in a hybrid Connectionist Temporal Classification (CTC)-Attention Conformer architecture. Additionally, we demonstrate further improvements through initialization of weights from a full-contextual model and parallelization of the convolution and self-attention modules. We evaluate our models on the open-source Voxpopuli, LibriSpeech and in-house conversational datasets. Overall, our proposed model reduces the degradation of the streaming mode over the non-streaming full-contextual model from 41.7% and 45.7% to 16.7% and 26.2% on the LibriSpeech test-clean and test-other datasets respectively, while improving by a relative 15.5% WER over the previous state-of-the-art unified model.
△ Less
Submitted 25 April, 2023; v1 submitted 18 April, 2023;
originally announced April 2023.
-
Investigation of octupole collectivity near the $A =72$ shape-transitional point
Authors:
M. Spieker,
L. A. Riley,
P. D. Cottle,
K. W. Kemper,
D. Bazin,
S. Biswas,
P. J. Farris,
A. Gade,
T. Ginter,
S. Giraud,
J. Li,
S. Noji,
J. Pereira,
M. Smith,
D. Weisshaar,
R. G. T. Zegers
Abstract:
Enhanced octupole collectivity is expected in the neutron-deficient Ge, Se and Kr isotopes with neutron number $N \approx 40$ and has indeed been observed for $^{70,72}$Ge. Shape coexistence and configuration mixing are, however, a notorious challenge for theoretical models trying to reliably predict octupole collectivity in this mass region, which is known to feature rapid shape changes with chan…
▽ More
Enhanced octupole collectivity is expected in the neutron-deficient Ge, Se and Kr isotopes with neutron number $N \approx 40$ and has indeed been observed for $^{70,72}$Ge. Shape coexistence and configuration mixing are, however, a notorious challenge for theoretical models trying to reliably predict octupole collectivity in this mass region, which is known to feature rapid shape changes with changing nucleon number and spin of the system. To further investigate the microscopic configurations causing the prolate-oblate-triaxial shape transition at $A \approx 72$ and their influence on octupole collectivity, the rare isotopes $^{72}$Se and $^{74,76}$Kr were studied via inelastic proton scattering in inverse kinematics. While significantly enhanced octupole strength of $\sim 32$ Weisskopf units (W.u.) was observed for $^{72}$Se, only strengths of $\sim 15$ W.u. were observed for $^{74,76}$Kr. In combination with existing data, the new data clearly question a simple origin of enhanced octupole strengths around $N = 40$. The present work establishes two regions of distinct octupole strengths with a sudden strength increase around the $A=72$ shape transitional point.
△ Less
Submitted 22 November, 2022;
originally announced November 2022.
-
Towards Personalization of CTC Speech Recognition Models with Contextual Adapters and Adaptive Boosting
Authors:
Saket Dingliwal,
Monica Sunkara,
Sravan Bodapati,
Srikanth Ronanki,
Jeff Farris,
Katrin Kirchhoff
Abstract:
End-to-end speech recognition models trained using joint Connectionist Temporal Classification (CTC)-Attention loss have gained popularity recently. In these models, a non-autoregressive CTC decoder is often used at inference time due to its speed and simplicity. However, such models are hard to personalize because of their conditional independence assumption that prevents output tokens from previ…
▽ More
End-to-end speech recognition models trained using joint Connectionist Temporal Classification (CTC)-Attention loss have gained popularity recently. In these models, a non-autoregressive CTC decoder is often used at inference time due to its speed and simplicity. However, such models are hard to personalize because of their conditional independence assumption that prevents output tokens from previous time steps to influence future predictions. To tackle this, we propose a novel two-way approach that first biases the encoder with attention over a predefined list of rare long-tail and out-of-vocabulary (OOV) words and then uses dynamic boosting and phone alignment network during decoding to further bias the subword predictions. We evaluate our approach on open-source VoxPopuli and in-house medical datasets to showcase a 60% improvement in F1 score on domain-specific rare words over a strong CTC baseline.
△ Less
Submitted 13 November, 2022; v1 submitted 17 October, 2022;
originally announced October 2022.
-
In-beam $γ$-ray spectroscopy at the proton dripline: $^{40}$Sc
Authors:
A. Gade,
D. Weisshaar,
B. A. Brown,
J. A. Tostevin,
D. Bazin,
K. Brown,
R. J. Charity,
P. J. Farris,
A. M. Hill,
J. Li,
B. Longfellow,
W. Reviol,
D. Rhodes
Abstract:
We report on the first in-beam $γ$-ray spectroscopy of the proton-dripline nucleus $^{40}$Sc using two-nucleon pickup onto an intermediate-energy rare-isotope beam of $^{38}$Ca. The $^{9}$Be($^{38}$Ca,$^{40}$Sc$+γ$)X reaction at 60.9 MeV/nucleon mid-target energy selectively populates states in $^{40}$Sc for which the transferred proton and neutron couple to high orbital angular momentum. In turn,…
▽ More
We report on the first in-beam $γ$-ray spectroscopy of the proton-dripline nucleus $^{40}$Sc using two-nucleon pickup onto an intermediate-energy rare-isotope beam of $^{38}$Ca. The $^{9}$Be($^{38}$Ca,$^{40}$Sc$+γ$)X reaction at 60.9 MeV/nucleon mid-target energy selectively populates states in $^{40}$Sc for which the transferred proton and neutron couple to high orbital angular momentum. In turn, due to angular-momentum selection rules in proton emission and the nuclear structure and energetics of $^{39}$Ca, such states in $^{40}$Sc then exhibit $γ$-decay branches although they are well above the proton separation energy. This work uniquely complements results from particle spectroscopy following charge-exchange reactions on $^{40}$Ca as well as $^{40}$Ti EC/$β^+$ decay which both display very different selectivities. The population and $γ$-ray decay of the previously known first $(5^-)$ state at 892 keV and the observation of a new level at 2744 keV are discussed in comparison to the mirror nucleus and shell-model calculations. On the experimental side, this work shows that high-resolution in-beam $γ$-ray spectroscopy is possible with new generation Ge arrays for reactions induced by rare-isotope beams on the level of a few $μ$b of cross section.
△ Less
Submitted 22 July, 2020;
originally announced July 2020.
-
Performance of Wavelength-Shifting Fibers for the Mu2e Cosmic Ray Veto Detector
Authors:
E. C. Dukes,
P. J. Farris,
R. C. Group,
T. Lam,
D. Shooltz,
Y. Oksuzian
Abstract:
The cosmic-ray-veto detector (CRV) for the Mu2e experiment consists of four layers of plastic scintillating counters read out by silicon photo-multipliers (SiPM) through wavelength-shifting fibers. This paper reports the light properties of several wavelength-shifting fiber samples with diameters of 1.0 mm, 1.4 mm, and 1.8 mm that were considered for the CRV system. A fiber diameter of 1.4 mm was…
▽ More
The cosmic-ray-veto detector (CRV) for the Mu2e experiment consists of four layers of plastic scintillating counters read out by silicon photo-multipliers (SiPM) through wavelength-shifting fibers. This paper reports the light properties of several wavelength-shifting fiber samples with diameters of 1.0 mm, 1.4 mm, and 1.8 mm that were considered for the CRV system. A fiber diameter of 1.4 mm was selected as optimal for the CRV, and measurements of the prototype and production fiber of this diameter are presented. Fiber performance was found to exceed the CRV requirements for >99% of the spools. The measurements are performed using a scanner designed to ensure the fiber quality for the CRV.
△ Less
Submitted 12 November, 2018;
originally announced November 2018.
