Reuse has been proposed as a microarchitecture-level mechanism to reduce the amount of executed instructions, collapsing dependencies and freeing resources for other instructions. Previous works have used reuse domain...
详细信息
Unmanned Aerial Vehicles (UAVs) are widely used in various applications, from inspection and surveillance to transportation and delivery. Navigating UAVs in complex 3D environments is a challenging task that requires ...
Unmanned Aerial Vehicles (UAVs) are widely used in various applications, from inspection and surveillance to transportation and delivery. Navigating UAVs in complex 3D environments is a challenging task that requires robust and efficient decision-making algorithms. This paper presents a novel approach to UAV navigation in 3D environments using a Curriculum-based Deep Reinforcement Learning (DRL) approach. The proposed method utilizes a deep neural network to model the UAV’s decision-making process and to learn a mapping from the state space to the action space. The learning process is guided by a reinforcement signal that reflects the performance of the UAV in terms of reaching its target while avoiding obstacles and with energy efficiency. Simulation results show that the proposed method has a positive trade off when compared to the baseline algorithm. The proposed method was able to perform well in environments with a state space size of 22 millions, allowing the usage in big environments or in maps with high resolution. The results demonstrate the potential of DRL for enabling UAVs to operate effectively in complex environments.
Most text-driven human motion generation methods employ sequential modeling approaches, e.g., transformer, to extract sentence-level text representations automatically and implicitly for human motion synthesis. Howeve...
Most text-driven human motion generation methods employ sequential modeling approaches, e.g., transformer, to extract sentence-level text representations automatically and implicitly for human motion synthesis. However, these compact text representations may overemphasize the action names at the expense of other important properties and lack fine-grained details to guide the synthesis of subtly distinct motion. In this paper, we propose hierarchical semantic graphs for fine-grained control over motion generation. Specifically, we disentangle motion descriptions into hierarchical semantic graphs including three levels of motions, actions, and specifics. Such global-to-local structures facilitate a comprehensive understanding of motion description and fine-grained control of motion generation. Correspondingly, to leverage the coarse-to-fine topology of hierarchical semantic graphs, we decompose the text-to-motion diffusion process into three semantic levels, which correspond to capturing the overall motion, local actions, and action specifics. Extensive experiments on two benchmark human motion datasets, including HumanML3D and KIT, with superior performances, justify the efficacy of our method. More encouragingly, by modifying the edge weights of hierarchical semantic graphs, our method can continuously refine the generated motion, which may have a far-reaching impact on the community. Code and pre-trained weights are available at https://***/jpthu17/GraphMotion.
In multi-converter power electronic systems, different converters such as DC/DC choppers, DC/AC inverters, and AC/DC rectifiers are used in source, load, and distribution subsystems to provide power at different volta...
详细信息
In multi-converter power electronic systems, different converters such as DC/DC choppers, DC/AC inverters, and AC/DC rectifiers are used in source, load, and distribution subsystems to provide power at different voltage levels and forms. Most of the loads are also in the form of power electronic converters and motor drives. The most popular examples of these systems are automotive systems and more electric/hybrid electric vehicles. These systems have unique characteristics, dynamics, and stability problems that are just beginning to be appreciated. In this paper, we take a closer look at multiconverter power electronic systems and address the fundamental problems faced in these systems.
The initial years of an infant’s life are known as the critical period, during which the overall development of learning performance is significantly impacted due to neural plasticity. In recent studies, an AI agent,...
详细信息
The creation of tools, techniques and methodologies to support the manipulation of large data sets has been receiving special attention of both scientific and industrial communities, in order to discover new ways of d...
详细信息
The creation of tools, techniques and methodologies to support the manipulation of large data sets has been receiving special attention of both scientific and industrial communities, in order to discover new ways of dealing with the underlying information, including learning purposes, identification of patterns, decision making support, amongst others. However, making use of computing resources to enhance awareness and understanding of software information and the software itself is still a challenge in software/systems engineering, since it involves the identification of suitable mechanisms, adequate abstractions, and studies on stimulation of the human perceptive and cognitive abilities. This paper presents some of the challenges in this context, based on current trends of software development lifecycle, program comprehension, and software engineering education. At the end, a special focus is given on ongoing research on using and improving current mechanisms for supporting software reuse practices and software comprehension in general.
Convolutional Neural Networks (CNN) have drawn the attention of researchers in the medical imaging field. Many researchers have exploited CNN for breast cancer detection. This study provides an Internet of Things (IoT...
Convolutional Neural Networks (CNN) have drawn the attention of researchers in the medical imaging field. Many researchers have exploited CNN for breast cancer detection. This study provides an Internet of Things (IoT) friendly implementation of CNN for breast cancer detection. To achieve faster time to Market, Deep-learning Processing Unit (DPU) on Field programmable Gate Array (FPGA) is adopted for the CNN hardware implementation. CNN inference on the proposed system achieves a 1.6x speed-up factor and 91.5% reduction in energy consumption compared to the conventional general-purpose multi-core Central Processing Unit (CPU).
Genes encoding early signaling events in pathogen defense often are identified only by their phenotype. Such genes involved in barley-powdery mildew interactions include Mla, specifying race-specific resistance; Rarl ...
详细信息
Genes encoding early signaling events in pathogen defense often are identified only by their phenotype. Such genes involved in barley-powdery mildew interactions include Mla, specifying race-specific resistance; Rarl （Required for Mla12-specified resistance1）, and Roml （Restoration of Mla-specified resistancel）. The HSP90-SGT1-RAR1 complex appears to function as chaperone in MLA-specified resistance, however, much remains to be discovered regarding the precise signaling underlying plant immunity. Genetic analyses of fast-neutron mutants derived from CI 16151 （Mla6） uncovered a novel locus, designated Rar3 （R_equired for Mla6-specified resitance3）. Rar3 segregates independent of Mla6 and Rarl, and rar3 mutants are susceptible to Blumeria graminis f. sp. hordei （Bgh） isolate 5874 （A VRar）, whereas, wild-type progenitor plants are resistant. Comparative expression analyses of the rar3 mutant vs. its wild-type progenitor were conducted via Barleyl GeneChip and GAIIx paired-end RNA-Seq. Whereas Rarl affects transcription of relatively few genes; Rar3 appears to influence thousands, notably in genes controlling ATP binding, catalytic activity, transcription, and phosphorylation; possibly membrane bound or in the nucleus, eQTL analysis of a segregating doubled haploid population identified over two-thousand genes as being regulated by Mla （q value/FDR=0.00001）, a subset of which are significant in Rar3 interactions. The intersection of datasets derived from mla-loss-of-function mutants, Mla-associated eQTL, and rar3-mediated transcriptome reprogramming are narrowing the focus on essential genes required for Mla-specified immunity.
In multi-converter power electronic systems, different power electronic converters are integrated together to form a complex and extensively interconnected system. In general, there are two kinds of loads in these sys...
详细信息
In multi-converter power electronic systems, different power electronic converters are integrated together to form a complex and extensively interconnected system. In general, there are two kinds of loads in these systems. One group is conventional loads, which have positive incremental impedance characteristics. They are mainly considered as constant voltage loads that require regulated voltage for their operation. The other group is tightly regulated power electronic converters and motor drives sinking constant power from their input buses. The purpose of this paper is to present an assessment of the dynamic interactions between these loads in the multi-converter DC power electronic systems. Furthermore, a nonlinear robust stabilizing controller based on the feedback linearization technique for DC/DC PWM converters is presented.
A capacitance humidity sensor is used as a test de- vice to characterize the performance of thirteen polyimide films in relative humidity sensing applications. This sensor has a multilayer, free-standing film construc...
详细信息
A capacitance humidity sensor is used as a test de- vice to characterize the performance of thirteen polyimide films in relative humidity sensing applications. This sensor has a multilayer, free-standing film construction. It consists of a humidity sensitive polyinfide dielectric core and conductive layers consisting of carbon filled polysul- fone on each side of the polyimide film to form a capacitor. Thirteen polyimide films, including commercial polyimides and films of novel chemistry, are investigated to determine the long term stability of sensors using the films exposed to 85'C/85% RH for a total of 28 days. Differences in film chemistry are used to interpret trends in the environmental stability of the films.
暂无评论