-
Federated Joint Learning of Robot Networks in Stroke Rehabilitation
Authors:
Xinyu Jiang,
Yibei Guo,
Mengsha Hu,
Ruoming **,
Hai Phan,
Jay Alberts,
Rui Liu
Abstract:
Advanced by rich perception and precise execution, robots possess immense potential to provide professional and customized rehabilitation exercises for patients with mobility impairments caused by strokes. Autonomous robotic rehabilitation significantly reduces human workloads in the long and tedious rehabilitation process. However, training a rehabilitation robot is challenging due to the data sc…
▽ More
Advanced by rich perception and precise execution, robots possess immense potential to provide professional and customized rehabilitation exercises for patients with mobility impairments caused by strokes. Autonomous robotic rehabilitation significantly reduces human workloads in the long and tedious rehabilitation process. However, training a rehabilitation robot is challenging due to the data scarcity issue. This challenge arises from privacy concerns (e.g., the risk of leaking private disease and identity information of patients) during clinical data access and usage. Data from various patients and hospitals cannot be shared for adequate robot training, further compromising rehabilitation safety and limiting implementation scopes. To address this challenge, this work developed a novel federated joint learning (FJL) method to jointly train robots across hospitals. FJL also adopted a long short-term memory network (LSTM)-Transformer learning mechanism to effectively explore the complex tempo-spatial relations among patient mobility conditions and robotic rehabilitation motions. To validate FJL's effectiveness in training a robot network, a clinic-simulation combined experiment was designed. Real rehabilitation exercise data from 200 patients with stroke diseases (upper limb hemiplegia, Parkinson's syndrome, and back pain syndrome) were adopted. Inversely driven by clinical data, 300,000 robotic rehabilitation guidances were simulated. FJL proved to be effective in joint rehabilitation learning, performing 20% - 30% better than baseline methods.
△ Less
Submitted 8 March, 2024;
originally announced March 2024.
-
Quantum state reduction of general initial states through spontaneous unitarity violation
Authors:
Aritro Mukherjee,
Srinivas Gotur,
Jelle Aalberts,
Rosa van den Ende,
Lotte Mertens,
Jasper van Wezel
Abstract:
The inability of Schrodinger's unitary time evolution to describe measurement of a quantum state remains a central foundational problem. It was recently suggested that the unitarity of Schrodinger dynamics can be spontaneously broken, resulting in measurement as an emergent phenomenon in the thermodynamic limit. Here, we introduce a family of models for spontaneous unitarity violation that apply t…
▽ More
The inability of Schrodinger's unitary time evolution to describe measurement of a quantum state remains a central foundational problem. It was recently suggested that the unitarity of Schrodinger dynamics can be spontaneously broken, resulting in measurement as an emergent phenomenon in the thermodynamic limit. Here, we introduce a family of models for spontaneous unitarity violation that apply to generic initial superpositions over arbitrarily many states, using either single or multiple state-independent stochastic components. Crucially, we show that Born's probability rule emerges spontaneously in all cases.
△ Less
Submitted 19 April, 2024; v1 submitted 9 January, 2023;
originally announced January 2023.
-
Precision constraints on radiative neutrino decay with CMB spectral distortion
Authors:
Jelle L. Aalberts,
Shin'ichiro Ando,
Wouter M. Borg,
Edwin Broeils,
Jennypher Broeils,
Stephen Broeils,
Bradley J. Kavanagh,
Gijs Leguijt,
Marnix Reemst,
Dylan R. van Arneman,
Hoang Vu
Abstract:
We investigate the radiative decay of the cosmic neutrino background, and its impact on the spectrum of the cosmic microwave background (CMB) that is known to be a nearly perfect black body. We derive exact formulae for the decay of a heavier neutrino into a lighter neutrino and a photon, $ν_j \to ν_i + γ$, and of absorption as its inverse, $ν_i + γ\to ν_j$, by accounting for the precise form of t…
▽ More
We investigate the radiative decay of the cosmic neutrino background, and its impact on the spectrum of the cosmic microwave background (CMB) that is known to be a nearly perfect black body. We derive exact formulae for the decay of a heavier neutrino into a lighter neutrino and a photon, $ν_j \to ν_i + γ$, and of absorption as its inverse, $ν_i + γ\to ν_j$, by accounting for the precise form of the neutrino momentum distribution. Our calculations show that if the neutrinos are heavier than $\mathcal O(0.1)$ eV, the exact formulae give results that differ by $\sim$50%, compared with approximate ones where neutrinos are assumed to be at rest. We also find that spectral distortion due to absorption is more important for heavy neutrino masses (by a factor of $\sim$10 going from a neutrino mass of 0.01 eV to 0.1 eV). By analyzing the CMB spectral data measured with COBE-FIRAS, we obtain lower limits on the neutrino lifetime of $τ_{12} \gtrsim 4 \times 10^{21}$ s (95% C.L.) for the smaller mass splitting and $τ_{13} \sim τ_{23} \gtrsim 10^{19}$ s for the larger mass splitting. These represent up to one order of magnitude improvement over previous CMB constraints. With future CMB experiments such as PIXIE, these limits will improve by roughly 4 orders of magnitude. This translates to a projected upper limit on the neutrino magnetic moment (for certain neutrino masses and decay modes) of $μ_ν< 3 \times 10^{-11}\, μ_B$, where $μ_B$ is the Bohr magneton. Such constraints would make future precision CMB measurements competitive with lab-based constraints on neutrino magnetic moments.
△ Less
Submitted 3 July, 2018; v1 submitted 1 March, 2018;
originally announced March 2018.
-
Force generation by Myosin II Filaments in Compliant Networks
Authors:
Samantha Stam,
Jon Alberts,
Margaret L. Gardel,
Edwin Munro
Abstract:
Myosin II isoforms with varying mechanochemistry and filament size interact with filamentous actin (F-actin) networks to generate contractile forces in cells. How their properties control force production in environments with varying stiffness is poorly understood. Here, we incorporated literature values for properties of myosin II isoforms into a cross-bridge model. Similar actin gliding speeds a…
▽ More
Myosin II isoforms with varying mechanochemistry and filament size interact with filamentous actin (F-actin) networks to generate contractile forces in cells. How their properties control force production in environments with varying stiffness is poorly understood. Here, we incorporated literature values for properties of myosin II isoforms into a cross-bridge model. Similar actin gliding speeds and force-velocity curves expected from previous experiments were observed. Motor force output on an elastic load was regulated by two timescales--that of their attachment to F-actin, which varied sharply with the ensemble size, motor duty ratio, and external load, and that of force build up, which scaled with ensemble stall force, gliding speed, and load stiffness. While such regulation did not require force-dependent kinetics, the myosin catch bond produced positive feedback between attachment time and force to trigger switch-like transitions from short attachments and small forces to high force-generating runs at threshold parameter values. Parameters representing skeletal muscle myosin, non-muscle myosin IIB, and non-muscle myosin IIA revealed distinct regimes of behavior respectively: (1) large assemblies of fast, low-duty ratio motors rapidly build stable forces over a large range of environmental stiffness, (2) ensembles of slow, high-duty ratio motors serve as high-affinity cross-links with force build-up times that exceed physiological timescales, and (3) small assemblies of low-duty ratio motors operating at intermediate speeds may respond sharply to changes in mechanical context--at low forces or stiffness, they serve as low affinity cross-links but they can transition to effective force production via the positive feedback mechanism described above. These results reveal how myosin isoform properties may be tuned to produce force and respond to mechanical cues in their environment.
△ Less
Submitted 8 July, 2014; v1 submitted 12 April, 2014;
originally announced April 2014.
