MRC Laboratory of Molecular Biology
CUED - Cambridge University Engineering DepartmentNot a member yet
45551 research outputs found
Sort by
Learning the electronic density of states in condensed matter
No full textThe electronic density of states (DOS) quantifies the distribution of the energy levels that can be occupied by electrons in a quasiparticle picture and is central to modern electronic structure theory. It also underpins the computation and interpretation of experimentally observable material properties such as optical absorption and electrical conductivity. We discuss the challenges inherent in the construction of a machine-learning (ML) framework aimed at predicting the DOS as a combination of local contributions that depend in turn on the geometric configuration of neighbors around each atom, using quasiparticle energy levels from density functional theory as training data. We present a challenging case study that includes configurations of silicon spanning a broad set of thermodynamic conditions, ranging from bulk structures to clusters and from semiconducting to metallic behavior. We compare different approaches to represent the DOS, and the accuracy of predicting quantities such as the Fermi level, the electron density at the Fermi level, or the band energy, either directly or as a side product of the evaluation of the DOS. We find that the performance of the model depends crucially on the resolution chosen to smooth the DOS and that there is a tradeoff to be made between the systematic error associated with the smoothing and the error in the ML model for a specific structure. We find however that the errors are not strongly correlated among similar structures, and so the average DOS over an ensemble of configurations is in very good agreement with the reference electronic structure calculations, despite the large nominal error on individual configurations. We demonstrate the usefulness of this approach by computing the density of states of a large amorphous silicon sample, for which it would be prohibitively expensive to compute the DOS by direct electronic structure calculations and show how the atom-centered decomposition of the DOS that is obtained through our model can be used to extract physical insights into the connections between structural and electronic features
Human-robot medical interaction
No full textAdvances in Soft Robotics, Haptics, AI and simulation have changed the medical robotics field, allowing robotics technologies to be deployed in medical environments. In this context, the relationship between doctors, robotics devices, and patients is fundamental, as only with the synergetic collaboration of the three parties results in medical robotics can be achieved. This workshop focuses on the use of soft robotics technologies, sensing, AI and Simulation, to further improve medical practitioner training, as well as the creation of new tools for diagnosis and healthcare through the medical interaction of humans and robots. The Robo-patient is more specifically the idea behind the creation of sensorised robotic patient with controllable organs to present a given set of physiological conditions. This is both to investigate the embodied nature of haptic interaction in physical examination, as well as the doctor-patient relationship to further improve medical practice through robotics technologies. The Robo-doctor aspect is also relevant, with robotics prototypes performing, or helping to perform, medical diagnosis. In the workshop, key technologies as well as future views in the field will be discussed both by expert and new upcoming researchers
Simulation study for magnetic levitation in pure water exploiting the ultra-high magnetic field gradient product of a hybrid trapped field magnet lens (HTFML)
No full textA hybrid trapped field magnet lens (HTFML) is a promising device that is able to concentrate a magnetic field higher than the applied field continuously, even after removing an external field, which was conceptually proposed by the authors in 2018. In this study, we propose a new additional advantage of the HTFML, which could be applicable for magnetic levitation and separation. The HTFML device consisting of a GdBaCuO bulk cylinder and a GdBaCuO magnetic lens, after the magnetization process from an applied field, Bapp = 10 T, can generate a maximum trapped field, Bc = 11.4 T, as well as an ultra-high magnetic field gradient product, BzdBz/dz, over ±3000 T2/m at Ts = 20 K, which is higher than that of existing superconducting magnets and large-scale hybrid magnets. Through detailed numerical simulations, the HTFML device is considered for the magnetic separation of a mixture of precious metal particles (Pt, Au, Ag, and Cu) dispersed in pure water, by exploiting the magneto-Archimedes effect. The HTFML can be realized as a compact and mobile desktop-type superconducting bulk magnet system, and there are a wide range of potential industrial applications, such as in the food and medical industries
Countering Acoustic Adversarial Attacks in Microphone-equipped Smart Home Devices
No full textDeep neural networks (DNNs) continue to demonstrate superior generalization performance in an increasing range of applications, including speech recognition and image understanding. Recent innovations in compression algorithms, design of efficient architectures and hardware accelerators have prompted a rapid growth in deploying DNNs on mobile and IoT devices to redefine user experiences. Relying on the superior inference quality of DNNs, various voice-enabled devices have started to pervade our everyday lives and are increasingly used for, e.g., opening and closing doors, starting or stopping washing machines, ordering products online, and authenticating monetary transactions. As the popularity of these voice-enabled services increases, so does their risk of being attacked. Recently, DNNs have been shown to be extremely brittle under adversarial attacks and people with malicious intentions can potentially exploit this vulnerability to compromise DNN-based voice-enabled systems. Although some existing work already highlights the vulnerability of audio models, very little is known of the behaviour of compressed on-device audio models under adversarial attacks. This paper bridges this gap by investigating thoroughly the vulnerabilities of compressed audio DNNs and makes a stride towards making compressed models robust. In particular, we propose a stochastic compression technique that generates compressed models with greater robustness to adversarial attacks. We present an extensive set of evaluations on adversarial vulnerability and robustness of DNNs in two diverse audio recognition tasks, while considering two popular attack algorithms: FGSM and PGD. We found that error rates of conventionally trained audio DNNs under attack can be as high as 100%. Under both white- and black-box attacks, our proposed approach is found to decrease the error rate of DNNs under attack by a large margin
Security assessment of urban areas through a gis-based analysis of lighting data generated by IoT sensors
No full textThe current perspective about urban development expects 70% of energy consumption will be concentrated in the cities in 2050. In addition, a growing density of people in the urban context leads to the need for increased security and safety for citizens, which imply a better lighting infrastructure. Smart solutions are required to optimize the corresponding energy effort. In developing countries, the cities' lighting is limited and the lighting world map is strongly significant about the urban density of the different areas. Nevertheless, in territories where the illumination level is particularly high, such as urban contexts, the conditions are not homogenous at the microscale level and the perceived security is affected by artificial urban lighting. As an example, 27.2% of the families living in the city of Milan, ombardy Region, Italy, consider critical the conditions of lighting in the city during the night, although the region has diffused infrastructure. The paper aims to provide a local illuminance geographic information system (GIS) mapping at the neighborhood level that can be extended to the urban context. Such an approach could unveil the need to increase lighting to enhance the perceived safety and security for the citizens and promote a higher quality of life in the smart city. Lighting mapping can be matched with car accident mapping of cities and could be extended to perceived security among pedestrians in urban roads and green areas, also related to degradation signs of the built environment. In addition, such an approach could open new scenarios to the adaptive street lighting control used to reduce the energy consumption in a smart city: the perceived security of an area could be used as an additional index to be considered during the modulation of the level of the luminosity of street lighting. An example of a measurement set-up is described and tested at the district level to define how to implement an extensive monitoring campaign based on an extended research schema
Fracture in distortion gradient plasticity
No full textIn this work, distortion gradient plasticity is used to gain insight into material deformation ahead of a crack tip. This also constitutes the first fracture mechanics analysis of gradient plasticity theories adopting Nye's tensor as primal kinematic variable. First, the asymptotic nature of crack tip fields is analytically investigated. We show that an inner elastic region exists, adjacent to the crack tip, where elastic strains dominate plastic strains and Cauchy stresses follow the linear elastic stress singularity. This finding is verified by detailed finite element analyses using a new numerical framework, which builds upon a viscoplastic constitutive law that enables capturing both rate-dependent and rate-independent behaviour in a computationally efficient manner. Numerical analysis is used to gain further insight into the stress elevation predicted by distortion gradient plasticity, relative to conventional J2 plasticity, and the influence of the plastic spin under both mode I and mixed-mode fracture conditions. It is found that Nye's tensor contributions have a weaker effect in elevating the stresses in the plastic region, while predicting the same asymptotic behaviour as constitutive choices based on the plastic strain gradient tensor. A minor sensitivity to X, the parameter governing the dissipation due to the plastic spin, is observed. Finally, distortion gradient plasticity and suitable higher order boundary conditions are used to appropriately model the phenomenon of brittle failure along elastic-plastic material interfaces. We reproduce paradigmatic experiments on niobium-sapphire interfaces and show that the combination of strain gradient hardening and dislocation blockage leads to interface crack tip stresses that are larger than the theoretical lattice strength, rationalising cleavage in the presence of plasticity at bi-material interfaces
Electrochemical impedance spectroscopy of human cochleas for modeling cochlear implant electrical stimulus spread
No full textCochlear implants (CIs) have tremendously helped people with severe to profound hearing loss to gain access to sound and oral-verbal communication. However, the electrical stimulus in the cochlea spreads easily and widely, since the perilymph and endolymph (i.e., intracochlear fluids) are essentially electrolytes, leading to an inability to focus stimulation to discrete portions of the auditory nerve, which blurs the neural signal. Here, we characterize the complex transimpedances of human cadaveric cochleas to investigate how electrical stimulus spread is distributed from 10 Hz to 100 kHz. By using electrochemical impedance spectroscopy (EIS), both the resistive and capacitive elements of human cochleas are measured and modeled with an electrical circuit model, identifying spread-induced and spread-independent impedance components. Based on this electrical circuit model, we implement a Laplace transform to simulate the theoretical shapes of the spread signals. The model is validated by experimentally applying the simulated stimulus as a real stimulus to the cochlea and measuring the shapes of the spread signals, with relative errors of <0.6% from the model. Based on this model, we show the relationship between stimulus pulse duration and electrical stimulus spread. This EIS technique to characterize the transimpedances of human cochleas provides a new way to predict the spread signal under an arbitrary electrical stimulus, thus providing preliminary guidance to the design of CI stimuli for different CI users and coding strategies
All-inkjet printed organic thin-film transistors with and without photo-sensitivity to visible lights
No full textPrintable organic thin-film transistors have enabled flexible low-cost electronics, which has the potential for a lot of emerging electronic applications. Despite the excellent dark performance of advanced all-inkjet printed organic thin-film transistors, their photoresponse is less explored and needs to be investigated, especially photoresponse to visible lights that human beings can see and are most familiar with. Importantly, for electronics integration, both devices with and without photo-sensitivity to visible light are important, for photo-detecting and signal processing, respectively. In this study, two organic semiconductor materials are used in all-inkjet printed organic thin-film transistors, namely 2,7-dioctyl[1]benzothieno[3,2-b][1]benzothiophene (C8-BTBT), 6,13-bis (triisopropylsilylethynyl) pentacene (TIPS-pentacene). By characterizing devices under optical exposure with wavelengths from 400 to 800 nm, photocurrents and threshold voltage shifts of the devices are extracted. The fabricated C8-BTBT organic thin-film transistors do not exhibit noticeable photo-sensitivity to visible light, whereas the TIPS-pentacene devices demonstrate significant photoresponse to visible lights, with photocurrents in nano-to micro-ampere levels and threshold voltage shifts of hundreds of millivolts to several volts depending on the photon energy of lights under the same intensity. The TIPS-pentacene devices demonstrated reproducible characteristics before and after light exposure. In addition, the responsivity and sensitivity of the devices were characterized with a decent responsivity of 55.9 mA/W. The photoresponse mechanisms are explained with ultraviolet–visible (UV–vis) adsorption spectroscopy measurements and extracted optical bandgaps of the two semiconductors. This study shows both printed organic transistors with and without photo-sensitivity can be fabricated with the same device structure and fabrication process at low cost, which opens the new possibility of using printed organic thin-film transistors for integrated optoelectronic applications
Angular dependence of secondary electron yield from microporous gold surfaces
No full textWe report exhaustive measurements of the secondary electron yield (SEY) from a gold film containing an array of micropores as a function of the angle of incidence of the primary electrons. The SEY measurements are in good agreement with Monte-Carlo (MC) simulations. A highly accurate empirical fit to the SEY data as a function of the incident electron impact angle is also proposed. In this study, the micropores have aspect ratios (ratio of pore height over pore diameter) ranging from about 1.5 to 3.5. The effect of the pore array density (porosity) and pore aspect ratio is analyzed in greater detail. It is found that increasing the pore aspect ratio and porosity leads to a sharp reduction in the total SEY in agreement with MC simulations
Addressing the need for standardization of test methods for self-healing concrete: an inter-laboratory study on concrete with macrocapsules.
No full textDevelopment and commercialization of self-healing concrete is hampered due to a lack of standardized test methods. Six inter-laboratory testing programs are being executed by the EU COST action SARCOS, each focusing on test methods for a specific self-healing technique. This paper reports on the comparison of tests for mortar and concrete specimens with polyurethane encapsulated in glass macrocapsules. First, the pre-cracking method was analysed: mortar specimens were cracked in a three-point bending test followed by an active crack width control technique to restrain the crack width up to a predefined value, while the concrete specimens were cracked in a three-point bending setup with a displacement-controlled loading system. Microscopic measurements showed that with the application of the active control technique almost all crack widths were within a narrow predefined range. Conversely, for the concrete specimens the variation on the crack width was higher. After pre-cracking, the self-healing effect was characterized via durability tests: the mortar specimens were tested in a water permeability test and the spread of the healing agent on the crack surfaces was determined, while the concrete specimens were subjected to two capillary water absorption tests, executed with a different type of waterproofing applied on the zone around the crack. The quality of the waterproofing was found to be important, as different results were obtained in each absorption test. For the permeability test, 4 out of 6 labs obtained a comparable flow rate for the reference specimens, yet all 6 labs obtained comparable sealing efficiencies, highlighting the potential for further standardization