-
Practical Trustworthiness Model for DNN in Dedicated 6G Application
Authors:
Anouar Nechi,
Ahmed Mahmoudi,
Christoph Herold,
Daniel Widmer,
Thomas Kürner,
Mladen Berekovic,
Saleh Mulhem
Abstract:
Artificial intelligence (AI) is considered an efficient response to several challenges facing 6G technology. However, AI still suffers from a huge trust issue due to its ambiguous way of making predictions. Therefore, there is a need for a method to evaluate the AI's trustworthiness in practice for future 6G applications. This paper presents a practical model to analyze the trustworthiness of AI i…
▽ More
Artificial intelligence (AI) is considered an efficient response to several challenges facing 6G technology. However, AI still suffers from a huge trust issue due to its ambiguous way of making predictions. Therefore, there is a need for a method to evaluate the AI's trustworthiness in practice for future 6G applications. This paper presents a practical model to analyze the trustworthiness of AI in a dedicated 6G application. In particular, we present two customized Deep Neural Networks (DNNs) to solve the Automatic Modulation Recognition (AMR) problem in Terahertz communications-based 6G technology. Then, a specific trustworthiness model and its attributes, namely data robustness, parameter sensitivity, and security covering adversarial examples, are introduced. The evaluation results indicate that the proposed trustworthiness attributes are crucial to evaluate the trustworthiness of DNN for this 6G application.
△ Less
Submitted 10 July, 2023;
originally announced July 2023.
-
Tuning Legged Locomotion Controllers via Safe Bayesian Optimization
Authors:
Daniel Widmer,
Dongho Kang,
Bhavya Sukhija,
Jonas Hübotter,
Andreas Krause,
Stelian Coros
Abstract:
This paper presents a data-driven strategy to streamline the deployment of model-based controllers in legged robotic hardware platforms. Our approach leverages a model-free safe learning algorithm to automate the tuning of control gains, addressing the mismatch between the simplified model used in the control formulation and the real system. This method substantially mitigates the risk of hazardou…
▽ More
This paper presents a data-driven strategy to streamline the deployment of model-based controllers in legged robotic hardware platforms. Our approach leverages a model-free safe learning algorithm to automate the tuning of control gains, addressing the mismatch between the simplified model used in the control formulation and the real system. This method substantially mitigates the risk of hazardous interactions with the robot by sample-efficiently optimizing parameters within a probably safe region. Additionally, we extend the applicability of our approach to incorporate the different gait parameters as contexts, leading to a safe, sample-efficient exploration algorithm capable of tuning a motion controller for diverse gait patterns. We validate our method through simulation and hardware experiments, where we demonstrate that the algorithm obtains superior performance on tuning a model-based motion controller for multiple gaits safely.
△ Less
Submitted 25 October, 2023; v1 submitted 12 June, 2023;
originally announced June 2023.
-
Nanometer scale resolution, multi-channel separation of spherical particles in a rocking ratchet with increasing barrier heights
Authors:
Philippe Nicollier,
Christian Schwemmer,
Francesca Ruggeri,
Daniel Widmer,
Xiaoyu Ma,
Armin W. Knoll
Abstract:
We present a nanoparticle size-separation device based on a nanofluidic rocking Brownian motor. It features a ratchet-shaped electrostatic particle potential with increasing barrier heights along the particle transport direction. The sharp drop of the particle current with barrier height is exploited to separate a particle suspension into multiple sub-populations. By solving the Fokker--Planck equ…
▽ More
We present a nanoparticle size-separation device based on a nanofluidic rocking Brownian motor. It features a ratchet-shaped electrostatic particle potential with increasing barrier heights along the particle transport direction. The sharp drop of the particle current with barrier height is exploited to separate a particle suspension into multiple sub-populations. By solving the Fokker--Planck equation, we show that the physics of the separation mechanism is governed by the energy landscape under forward tilt of the ratchet. For a given device geometry and sorting duration, the applied force is thus the only tunable parameter to increase the separation resolution. For the experimental conditions of 3.5 V applied voltage and 20 s sorting, we predict a separation resolution of $\sim 2$ nm, supported by experimental data for separating spherical gold particles of nominal 80 and 100 nm diameters.
△ Less
Submitted 8 July, 2020;
originally announced July 2020.
-
CMOS Platform for Atomic-Scale Device Fabrication
Authors:
Tomas Skeren,
Nikola Pascher,
Arnaud Garnier,
Patrick Reynaud,
Emmanuel Rolland,
Aurelie Thuaire,
Daniel Widmer,
Xavier Jehl,
Andreas Fuhrer
Abstract:
Controlled atomic scale fabrication of functional devices is one of the holy grails of nanotechnology. The most promising class of techniques that enable deterministic nanodevice fabrication are based on scanning probe patterning or surface assembly. However, this typically involves a complex process flow, stringent requirements for an ultra high vacuum environment, long fabrication times and, con…
▽ More
Controlled atomic scale fabrication of functional devices is one of the holy grails of nanotechnology. The most promising class of techniques that enable deterministic nanodevice fabrication are based on scanning probe patterning or surface assembly. However, this typically involves a complex process flow, stringent requirements for an ultra high vacuum environment, long fabrication times and, consequently, limited throughput and device yield. Here, a device platform is developed that overcomes these limitations by integrating scanning probe based dopant device fabrication with a CMOS-compatible process flow. Silicon on insulator substrates are used featuring a reconstructed Si(001):H surface that is protected by a cap** chip and has pre-implanted contacts ready for scanning tunneling microscope (STM) patterning. Processing in ultra-high vacuum is thus reduced to only a few critical steps which minimizes the complexity, time and effort required for fabrication of the nanoscale dopant devices. Subsequent reintegration of the samples into the CMOS process flow not only simplifies the post-processing but also opens the door to successful application of STM based dopant devices as a building block in more complex device architectures. Full functionality of this approach is demonstrated with magnetotransport measurements on degenerately doped STM patterned Si:P nanowires up to room temperature.
△ Less
Submitted 1 October, 2019;
originally announced October 2019.
-
Atomic structure of Mn wires on Si(001) resolved by scanning tunneling microscopy
Authors:
A. Fuhrer,
F. J. Rueß,
N. Moll,
A. Curioni,
D. Widmer
Abstract:
At submonolayer coverage, Mn forms atomic wires on the Si(001) surface oriented perpendicular to the underlying Si dimer rows. While many other elements form symmetric dimer wires at room temperature, we show that Mn wires have an asymmetric appearance and pin the Si dimers nearby. We find that an atomic configuration with a Mn trimer unit cell can explain these observations due to the interplay b…
▽ More
At submonolayer coverage, Mn forms atomic wires on the Si(001) surface oriented perpendicular to the underlying Si dimer rows. While many other elements form symmetric dimer wires at room temperature, we show that Mn wires have an asymmetric appearance and pin the Si dimers nearby. We find that an atomic configuration with a Mn trimer unit cell can explain these observations due to the interplay between the Si dimer buckling phase near the wire and the orientation of the Mn trimer. We study the resulting four wire configurations in detail using high-resolution scanning tunneling microscopy (STM) imaging and compare our findings with STM images simulated by density functional theory.
△ Less
Submitted 2 May, 2012;
originally announced May 2012.
-
Electrical current distribution across a metal-insulator-metal structure during bistable switching
Authors:
C. Rossel,
G. I. Meijer,
D. Bremaud,
D. Widmer
Abstract:
Combining scanning electron microscopy (SEM) and electron-beam-induced current (EBIC) imaging with transport measurements, it is shown that the current flowing across a two-terminal oxide-based capacitor-like structure is preferentially confined in areas localized at defects. As the thin-film device switches between two different resistance states, the distribution and intensity of the current p…
▽ More
Combining scanning electron microscopy (SEM) and electron-beam-induced current (EBIC) imaging with transport measurements, it is shown that the current flowing across a two-terminal oxide-based capacitor-like structure is preferentially confined in areas localized at defects. As the thin-film device switches between two different resistance states, the distribution and intensity of the current paths, appearing as bright spots, change. This implies that switching and memory effects are mainly determined by the conducting properties along such paths. A model based on the storage and release of charge carriers within the insulator seems adequate to explain the observed memory effect.
△ Less
Submitted 24 April, 2001;
originally announced April 2001.
