196 research outputs found
Recovery from Non-Decomposable Distance Oracles
Get PDFA line of work has looked at the problem of recovering an input from distance
queries. In this setting, there is an unknown sequence , and one chooses a set of queries and
receives for a distance function . The goal is to make as few
queries as possible to recover . Although this problem is well-studied for
decomposable distances, i.e., distances of the form for some function , which includes the important cases of
Hamming distance, -norms, and -estimators, to the best of our
knowledge this problem has not been studied for non-decomposable distances, for
which there are important special cases such as edit distance, dynamic time
warping (DTW), Frechet distance, earth mover's distance, and so on. We initiate
the study and develop a general framework for such distances. Interestingly,
for some distances such as DTW or Frechet, exact recovery of the sequence
is provably impossible, and so we show by allowing the characters in to be
drawn from a slightly larger alphabet this then becomes possible. In a number
of cases we obtain optimal or near-optimal query complexity. We also study the
role of adaptivity for a number of different distance functions. One motivation
for understanding non-adaptivity is that the query sequence can be fixed and
the distances of the input to the queries provide a non-linear embedding of the
input, which can be used in downstream applications involving, e.g., neural
networks for natural language processing.Comment: This work has been presented at conference The 14th Innovations in
Theoretical Computer Science (ITCS 2023) and accepted for publishing in the
journal IEEE Transactions on Information Theor
Long Song Lyrics (manci) of the Song Dynasty : Li Qingzhao : Singing her autumn sorrow
Get PDFA master of tune and sense, Li Qingzhao knows how to bring out her almost unspeakable inner feeling through her skillful employment of the ci form, the music of words
Poly(3,4-ethylenedioxythiophene) (PEDOT) Coatings for High-Quality Electromyography Recording
No full textConducting polymer coatings on metal electrodes are an efficient solution to improve neural signal recording and stimulation, due to their mixed electronic-ionic conduction and biocompatibility. To date, only a few studies have been reported on conducting polymer coatings on metallic wire electrodes for muscle signal recording. Chronic muscle signal recording of freely moving animals can be challenging to acquire with coated electrodes, due to muscle movement around the electrode that can increase instances of coating delamination and device failure. The poor adhesion of conducting polymers to some inorganic substrates and the possible degradation of their electrochemical properties after harsh treatments, such as sterilization, or during implantation limits their use for biomedical applications. Here, we demonstrate the mechanical and electrochemical stability of the conducting polymer, poly(3,4-ethylenedioxythiophene) (PEDOT) doped with LiClO4, deposited on stainless steel multistranded wire electrodes for invasive muscle signal recording in mice. The mechanical and electrochemical stability was achieved by tuning the electropolymerization conditions. PEDOT-coated and bare stainless steel electrodes were implanted in the neck muscle of five mice for electromyographic (EMG) activity recording over a period of 6 weeks. The PEDOT coating improved the electrochemical properties of the stainless steel electrodes, lowering the impedance, resulting in an enhanced signal-to-noise ratio during in vivo EMG recording compared to bare electrodes
TACOformer:Token-channel compounded Cross Attention for Multimodal Emotion Recognition
Get PDFRecently, emotion recognition based on physiological signals has emerged as a
field with intensive research. The utilization of multi-modal, multi-channel
physiological signals has significantly improved the performance of emotion
recognition systems, due to their complementarity. However, effectively
integrating emotion-related semantic information from different modalities and
capturing inter-modal dependencies remains a challenging issue. Many existing
multimodal fusion methods ignore either token-to-token or channel-to-channel
correlations of multichannel signals from different modalities, which limits
the classification capability of the models to some extent. In this paper, we
propose a comprehensive perspective of multimodal fusion that integrates
channel-level and token-level cross-modal interactions. Specifically, we
introduce a unified cross attention module called Token-chAnnel COmpound (TACO)
Cross Attention to perform multimodal fusion, which simultaneously models
channel-level and token-level dependencies between modalities. Additionally, we
propose a 2D position encoding method to preserve information about the spatial
distribution of EEG signal channels, then we use two transformer encoders ahead
of the fusion module to capture long-term temporal dependencies from the EEG
signal and the peripheral physiological signal, respectively.
Subject-independent experiments on emotional dataset DEAP and Dreamer
demonstrate that the proposed model achieves state-of-the-art performance.Comment: Accepted by IJCAI 2023- AI4TS worksho
Simulation of stably controlled biped walking by educational humanoid robot
No full textThe study of humanoid robots has become an important direction in the research of
robotics, and has been extensively applied across various areas such as the service
industry and healthcare. Among these applications, the “humanoid” features of
humanoid robots give them greater affinity and interactivity in the educational
applications than other robots. Given that educational robots are mainly designed for
children and students, research on the motion stability and safety is very important.
In this case, this dissertation reviews the development history of humanoid robot
technology and investigates into several theoretical methods of stability control for
biped walking robots. Based on the Linear Inverted Pendulum Model (LIPM), this
dissertation conducts a walking simulation of a biped model in MATLAB using the
theory of orbital energy. The simulation analyzes the impact of various walking
parameters on the walking control of biped walking robots and explores the
approaches to achieve a stable walking process.Master's degre
Structure–Property Relations of Polyampholyte Hydrogels, Graphene Composites, and Biodegradable Elastomers and Their Applications in Energy Generation and Storage Devices
No full textA polyampholyte is a polyelectrolyte which is composed of macromolecules that contains both cationic and anionic ionisable groups. Understanding the structure of polymer chains in polyampholyte hydrogels and its effect on the phase behavior of the contained water is a challenging problem that has a broad impact on diverse applications, such as gel electrolytes, lubrication layers, and anti-biofouling coatings. In this dissertation, we have investigated the structure of a charge-balanced polyampholyte, poly(4-vinylbenzenesulfonate-co-[3-(methacryloylamino) propyl] trimethylammonium chloride). The polymer structure was probed by variable-temperature small-angle X-ray scattering (SAXS); highly hydrated globules with a radius of gyration of 2 ~ 2.5 nm formed a network structure in the charge-balanced polyampholyte hydrogels, whereas the size and the clustering are dependent on synthesis parameters. These highly hydrated network globules formed percolated polymer-rich domains while sub-micron, slush-like ice crystals formed from water-rich domains, resulting in ion-conducting channels that are rich in amorphous water molecules at low temperatures. Solid-state nuclear magnetic resonance (NMR) spectroscopy confirmed the mobility of these amorphous water molecules at temperatures as low as –54 C. We also visualized the globular structure of polyampholyte hydrogel with scanning electron microscopy (SEM) and scanning transmission electron microscopy (STEM) for the first time. Differential scanning calorimetry (DSC) was used to estimate water states in the polyampholyte hydrogel.
Based on the understanding of temperature-dependent structure evolution of polyampholyte hydrogel, we fabricated two devices using polyampholyte hydrogel. First, we found that the upper critical solution temperature (UCST) for the phase separation between water-rich and polymer-rich phases in the polyampholyte hydrogel can be finely raised (~ +40 °C) by substituting a small number of cationic monomers (~ 0.5 wt%) to slightly more hydrophobic ones during the random copolymerization process. By harnessing this scientific observation, we developed a smart coating that becomes opaque to both visible and mid-infrared radiation at room temperature to achieve privacy and heat retention in cold night weather. Second, a flexible and self-healing supercapacitor with high energy density in low-temperature operation was fabricated using the polyampholyte hydrogel as a gel electrolyte. The resulting supercapacitor device showed a high energy density of 30 Wh/kg, and a capacitance retention of ~90% after 5000 charge–discharge cycles. At low temperature (−30 °C), the supercapacitor had an energy density of 10.5 Wh/kg at a power density of 500 W/kg.
Inspired by the new graphene chemistry we developed for the aforementioned supercapacitor electrodes, we also devised a self-reinforcing conductive coating strategy where graphene nanoflakes (GNF) were wrapped by self-assembling reduced graphene oxide (RGO) for electrical conductivity and mechanical integrity. The conductivity of the GNF-RGO coating reached 4.47 × 104 S/m. The coating was then applied on 3D printed porous elastomers, resulting in flexible radio frequency (RF) antennas and strain sensors with arbitrary shapes for internet-of-things (IoTs) applications. The same conductive coating strategy also converted a commercial polyurethane (PU) sponge into electricity generators, where zinc oxide (ZnO) nanowires were hydrothermally grown on top of the three-dimensional graphene networks coated on the inner wall of the PU sponge. The nanogenerator yielded an open circuit voltage of ∼0.5 V and short circuit current density of ∼2 μA/cm2, while the output was found to be consistent after ∼3000 cycles.
Finally, we studied the synthesis of poly (glycerol sebacate) (PGS), a synthetic biocompatible elastomer developed as a template for cardiac cells, to establish criteria for quick and consistent synthesis for tailored mechanical properties. Here, we suggested that the degree of esterification (DE) could be used to predict precisely the physical status, the mechanical properties, and the degradation of PGS. Young’s modulus was shown to linearly increase with DE, which was in agreement with an entropic spring theory of rubbers. To provide a processing guideline for researchers, we also provided a physical status map as a function of curing temperature and time. The amount of glycerol loss, obtainable by monitoring the evolution of the total mass loss and the DE during synthesis, was shown to make the predictions even more precise. Quick synthesis could be achieved by employing a microwave oven instead of convection heating for prepolymerization, but microwave heating led to severe mismatch between monomeric units by uncontrolled evaporation of glycerol, leading to an inconsistency in mechanical properties
Improved Model Poisoning Attacks and Defenses in Federated Learning with Clustering
Get PDFFederated Learning (FL) allows multiple participants to collaboratively train a deep learning model without sharing their private training data. However, due to its distributive nature, FL is vulnerable to various poisoning attacks. An adversary can submit malicious model updates that aim to degrade the joint model's utility. In this thesis, we formulate the adversary's goal as an optimization problem and present an effective model poisoning attack using projected gradient descent. Our empirical results show that our attack has a larger impact on the global model's accuracy than previous attacks.
Motivated by this, we design a robust defense algorithm that mitigates existing poisoning attacks. Our defense leverages constraint k-means clustering and uses a small validation dataset for the server to select optimal updates in each FL round. We conduct experiments on three non-iid image classification datasets and demonstrate the robustness of our defense algorithm under various FL settings
The Importance of Keratinized Mucosa Around Dental Implants
Get PDFAIM: The objective of this review is to determine the efficiency of different augmentation materials in increasing the ideal width and thickness of peri-implant keratinized mucosa (PIKM) and thereby achieving stable peri-implant health. To this end, various clinical parameters will be analyzed for the different techniques employed and the initially achieved PIKM width and thickness.
METHODS: An electronic search of the literature from 2009 to 2019 was conducted in PubMed for studies addressing peri-implant keratinized mucosa and peri-implant health, in combination with peri-implant soft tissue augmentation.
RESULTS: The initial search identified 311 studies, of which 28 passed the first review phase, resulting in 19 studies selected based on established criteria. Insufficient peri-implant keratinized mucosa width (< 2 mm) correlates with inflammation. The combination of autogenous grafts (free gingival grafts) and an apically positioned flap is the gold standard for gaining PIKM width, while xenogeneic materials may offer a short-term alternative. For increasing PIKM thickness, Alloderm, Mucoderm, or autogenous grafts used in augmentation procedures significantly reduced marginal bone loss.
CONCLUSION: PIKM width under 2 mm should be avoided due to its association with increased marginal bone loss and inflammation. Autograft augmentation combined with an apically positioned flap is the gold standard, while allogenic and xenogenic materials also yield successful outcomes
Efficacy of Augmentation Materials and Surgical Methods in Alveolar Ridge Preservation Post-Tooth Extraction
Get PDFAIM: To evaluate the clinical and radiographic outcomes of various augmentation materials and surgical techniques in alveolar ridge preservation procedures compared to natural post-extraction socket healing.
METHODS: A literature search was conducted using the PubMed/MEDLINE database, focusing on the efficacy of different augmentation materials. The initial search yielded 291 studies, which were screened to exclude duplicates and irrelevant entries. Full texts of potentially relevant articles were assessed based on predefined inclusion criteria: studies had to be randomized controlled trials or clinical studies with a minimum follow-up of two months.
RESULTS: In total, 19 studies were ultimately included into the analysis. Alveolar ridge preservation (ARP) procedures are more effective than natural healing in minimizing post-extraction bone resorption, preserving both horizontal and vertical bone dimensions. Cortico-cancellous porcine bone particles and alloplastic materials yield superior results in maintaining alveolar ridge dimensions compared to control groups. Platelet-rich fibrin also reduces bone resorption and enhances preservation. Barrier membranes in ARP procedures further improve outcomes.
CONCLUSION: Future research should refine ARP techniques and materials, focusing on long-term effectiveness and practicality. Investigations into cost-effectiveness and ease of application will promote broader adoption. By implementing tailored ARP strategies, dental professionals can enhance the long-term success of restorations, sustaining patients\u27 health, function, and aesthetics
Ultrastrong and Tough Urushiol-Based Ionic Conductive Double Network Hydrogels as Flexible Strain Sensors
No full textIonic conductive hydrogels have attracted increasing research interest in flexible electronics. However, the limited resilience and poor fatigue resistance of current ionic hydrogels significantly restrict their practical application. Herein, an urushiol-based ionic conductive double network hydrogel (PU/PVA-Li) was developed by one-pot thermal initiation polymerization assisted with freeze–thaw cycling and subsequent LiCl soaking. Such a PU/PVA-Li hydrogel comprises a primary network of covalently crosslinked polyurushiol (PU) and a secondary network formed by physically crosslinked poly(vinyl alcohol) (PVA) through crystalline regions. The obtained PU/PVA-Li hydrogel demonstrates exceptional mechanical properties, including ultrahigh strength (up to 3.4 MPa), remarkable toughness (up to 1868.6 kJ/m3), and outstanding fatigue resistance, which can be attributed to the synergistic effect of the interpenetrating network structure and dynamic physical interactions between PU and PVA chains. Moreover, the incorporation of LiCl into the hydrogels induces polymer chain contraction via ionic coordination, further enhancing their mechanical strength and resilience, which also impart exceptional ionic conductivity (2.62 mS/m) to the hydrogels. Based on these excellent characteristics of PU/PVA-Li hydrogel, a high-performance flexible strain sensor is developed, which exhibits high sensitivity, excellent stability, and reliability. This PU/PVA-Li hydrogel sensor can be effectively utilized as a wearable electronic device for monitoring various human joint movements. This PU/PVA-Li hydrogel sensor could also demonstrate its great potential in information encryption and decryption through Morse code. This work provides a facile strategy for designing versatile, ultrastrong, and tough ionic conductive hydrogels using sustainable natural extracts and biocompatible polymers. The developed hydrogels hold great potential as promising candidate materials for future flexible intelligent electronics
- …
