教学课程资源库 更多>>

关键词： Computer science   Artificial intelligence  

关键词： Biomedical engineering   Electrical engineering   Computer science   Neurosciences   Artificial intelligence  

摘要： Objective. A symbiosis of human intelligence and artificial intelligence (AI) cannot be achieved without establishing an intuitive, bidirectional, and high-bandwidth information conduit between the minds and machines. Approach. Here we focus on developing high-precision bioelectronics underlying a new class of bioelectric neural interfaces that could bring us one step closer to this feat. We pioneer new circuit techniques, including frequency shaping (FS), redundant sensing (RS), RS-based super-resolution, and redundant crossfire (RXF), to enhance the effective resolution of neural recording and stimulation. These fundamentals allow the implementation of a series of fully-integrated microchips called Neuronix capable of acquiring low-noise neural signals and delivering high-precision electrical microstimulation. The Neuronix chips are incorporated to create miniaturized neuromodulation devices, including the Scorpius system, to enable bidirectional communications with neural circuits. Results. In a clinical study with human amputees, the Scorpius system helps establish a peripheral nerve interface that allows deep learning-based AI models to read and decode the patients' intents of moving individual fingers. Our analysis of acquired electroneurography (ENG) signals demonstrates this robust nerve interface has a sufficient information capacity to enable real-time control of a multi-degree-of-freedom (DOF) neuroprosthetic hand with near-natural dexterity and intuitiveness while simultaneously delivering somatosensory feedback. Significance. Our study layouts the principled foundation toward not only a dexterous control strategy for advanced neuroprostheses but also an intuitive conduit for connecting the human minds and machines. This opens up possibilities for many biomedical applications and manifests the basis of the future human-machine symbiosis.

关键词： Computer science   Sedimentary Geology   Geology   Hydraulic engineering   Physical geography  

摘要： Floods are the most impactful natural disasters on earth, and reliable flood warning systems are critical for disaster preparation, mitigation, and response. The GEOGloWS ECMWF Streamflow Services (GESS) provide forecasted streamflow throughout the world. While forecasted discharge is essential to flood warning, forecasted inundation extents are required to understand and predict flood impact. In this research, I sought to expand GESS flood warning potential by generating inundation extents from streamflow forecasts. I compared Height Above Nearest Drainage (HAND), a method beneficial for flood mapping on a watershed scale, to a 2D hydrodynamic model, specifically Sedimentation and River Hydraulics – Two Dimension (SRH-2D), a method localized to specific areas of high importance. In three study areas in the Amazon basin, I validated HAND and SRH-2D flood maps against water maps derived from satellite SAR imagery. Specifically, I analyzed what features of an SRH-2D model were required to generate more accurate flood extents than HAND. I also analyzed the practicality of using SRH-2D for forecasting by comparing flood extents generated from simulating a complete forecast hydrograph to flood extents precomputed at predetermined, incremental flowrates. The SRH-2D models outperformed HAND, but their accuracy decreased at flowrates different than those used for calibration, limiting their reliability for forecasting and impact analysis. Based on this study, the key features necessary for a reliable SRH-2D model for forecasting include (1) a high-resolution DEM for an accurate representation of the floodplain, (2) correct representation of channel flow control, and (3) a channel bathymetry approximation and exit boundary rating curve that correctly predict water levels at a range of input flowrates. For forecasting practicality, the precomputed flood extents had accuracies comparable to the complete hydrograph simulations, showing their potential for estimating forecaste

关键词： Computer science   Aerospace engineering   Artificial intelligence  

关键词： Lesions   Mammography   Data models   Generative adversarial networks   Estimation   Computer science   Breast cancer   Domain Knowledge   Bilateral Asymmetry   Counterfactual   Mammogram Classification  

摘要： Mammogram benign or malignant classification with only image-level labels is challenging due to the absence of lesion annotations. Motivated by the symmetric prior that the lesions on one side of breasts rarely appear in the corresponding areas on the other side, we explore to answer a counterfactual question to identify the lesion areas. This counterfactual question means: given an image with lesions, how would the features have behaved if there were no lesions in the image? To answer this question, we derive a new theoretical result based on the symmetric prior. Specifically, by building a causal model that entails such a prior for bilateral images, we identify to optimize the distances in distribution between i) the counterfactual features and the target side's features in lesion-free areas;and ii) the counterfactual features and the reference side's features in lesion areas. To realize these optimizations for better benign/malignant classification, we propose a counterfactual generative network, which is mainly composed of Generator Adversarial Network and a prediction feedback mechanism, they are optimized jointly and prompt each other. Specifically, the former can further improve the classi?cation performance by generating counterfactual features to calculate lesion areas. On the other hand, the latter helps counterfactual generation by the supervision of classification loss. The utility of our method and the effectiveness of each module in our model can be verified by state-of-the-art performance on INBreast and an in-house dataset and ablation studies.

关键词： Computer science   Computer engineering   Engineering  

摘要： Reliable detection of human occupancy in indoor environments is critical for various energy efficiency, security and safety applications. We consider this challenge of occupancy detection using extremely low quality, privacy-preserving images from low power image sensors. We propose a combined few shot learning and clustering algorithm to address this challenge that has very low commissioning and maintenance cost. While the few shot learning concept enables us to commission our system with a few labeled examples, the clustering step serves the purpose of online adaptation to changing imaging environment over time. Apart from validating and comparing our algorithm on benchmark datasets, we also demonstrate performance of our algorithm on streaming images collected from real homes using our novel battery free camera hardware that also leads to a new dataset for the vision community

关键词： Computer science   Artificial intelligence   Datasets   Poisoning   Privacy   Neural networks  

摘要： Machine learning is being used for an increasing number of applications with societal impact. In such settings, models must be trusted to be fair, useful, and robust. In many applications, a large amount of training data is collected from a variety of sources, including from private or untrusted individuals. Manually sanitizing datasets is difficult as datasets increase in size, which allows a motivate adversary to adversarially corrupt training data, crafted to impact the model at test time, referred to as a poisoning attack. These poisoning attacks can cause specific users’ data to be misclassified, which can be harmful if models are applied to sensitive tasks such as security applications. At the same time, models trained on datasets collected from real people must protect their privacy, preventing unscrupulous onlookers from learning more than they should; in sensitive domains such as personalized medicine, privacy is of utmost importance. In this thesis, we describe integrity and privacy vulnerabilities in these critical settings. We explore the variety of adversarial goals that can be accomplished with poisoning, and how to construct defenses against these attacks (if this is possible at all). Our work will high-light the difficulty of developing generic poisoning defenses; dependence on the adversarial objective appears to be necessary for large enough attacks. Next, we discuss the connection between differential privacy and poisoning attacks, showing that poisoning can be useful for interpreting privacy guarantees, and differential privacy may not serve as a defense from poisoning attacks. Finally, we discuss privacy leakage in the realistic training setting where models are updated repeatedly over time. Our privacy attacks highlight and tackle the current challenges in deploying private algorithms in real world settings. Overall, this thesis will demonstrate the diversity of types of both poisoning attacks and privacy attacks and the challenges in defending

关键词： Mechanical engineering   Materials science   Computer science  

摘要： The utilization of additive manufacturing (AM) is rising tremendously in every aspect of our daily life. One of the most widely utilized AM technologies is Fused Filament Fabrication (FFF). Fortunately, this technique has several advantages in fabricating low-cost and functional products. With the advancement of filaments made with several different materials, the applications of FFF have grown in a high number of industries. Understanding the impact of various printing parameters on the mechanical properties of FFF-made parts is an important research question to investigate. Predicting the mechanical behaviors of the fabricated components will help the engineers, researchers, and end-users easily access and use the findings of this study. Although several investigations reported the tensile and compression behaviors of short carbon fiber reinforced FFF-made products, there is still a need to know the bending behaviors for real-life applications. It is also important to predict the future values of the bending properties so the end users can easily implement the findings of these studies for their needs. This thesis reports the application of machine learning (ML) based extrapolative models for the prediction of flexural properties of additively manufactured composite parts. The ability to predict flexural properties with high accuracy portrays an important contribution to the knowledge base of fiber-reinforced AM. In our case, ML models provide insights into the flexural behaviors of FFF-made parts. Overall, this research work creates knowledge blocks about the accurate prediction of flexural properties of short carbon fiber FFF-made parts.

关键词： Computer engineering   Computer science   Web studies   Artificial intelligence  

关键词： Mechanics   Physiology   Biology   Computer science   Morphology   Optics   Physics  

摘要： Smooth muscle has the unique ability to maintain force for long periods of time at low energy cost. This characteristic is called the latch-state and occurs more frequently in tonic than phasic smooth muscle. Although many theories have been proposed to explain the latch-state, all were based on tissue level measurements and have yet to be verified at the molecular level. One such theory suggests that, if myosin is dephosphorylated while attached to actin, it will remain attached and maintain force. To elucidate the underlying mechanism of the latch-state, myosin light chain phosphatase (MLCP) was injected in the laser trap assay while measuring the force generated by a population of tonic or phasic smooth muscle myosin (SMM) molecules pulling on a single actin filament. To further understand the effect of dephosphorylation on the SMM molecular mechanics, MLCP was also injected during in vitro motility assays consisting of 100% tonic or phasic SMM and in vitro motility mixture assays consisting of 25% skeletal muscle myosin (SKM) and 75% tonic or phasic SMM, while measuring the velocity (v) and fraction of moving actin filaments (fmot). The in vitro motility assays provided information regarding the time course of SMM deactivation, whereas the rationale behind the mixture assays was that, since SKM is not regulated by phosphorylation, a transient decrease in v and/or fmot due to the load imposed by attached, dephosphorylated SMM (latch-bridges) should be observed if the latch-state occurs. Additionally, v and/or fmot should eventually increase to the level of SKM following latch-bridge detachment. Flow-through chambers with a top opening covered by a microporous membrane to allow diffusion of MLCP without creating bulk flow were optimized and improved the quality of molecular mechanics measurements. Force maintenance post MLCP injection was statistically longer for tonic than phasic SMM. The rate at which fmot decreased during the in vitro motility assays was not st