With an ultra-compact footprint of 9×18μm2 on-chip diffractive optical logic operation units accomplish (N)AND, (N)OR, and X(N)OR utilizing a standardized structures with only 36 etching slots. It has outstandin...
详细信息
Today's deep learning models face an increasing demand to handle dynamic shape tensors and computation whose shape information remains unknown at compile time and varies in a nearly infinite range at runtime. This...
详细信息
Today's deep learning models face an increasing demand to handle dynamic shape tensors and computation whose shape information remains unknown at compile time and varies in a nearly infinite range at runtime. This shape dynamism brings tremendous challenges for existing compilation pipelines designed for static models which optimize tensor programs relying on exact shape values. This paper presents TSCompiler, an end-to-end compilation framework for dynamic shape models. TSCompiler first proposes a symbolic shape propagation algorithm to recover symbolic shape information at compile time to enable subsequent optimizations. TSCompiler then partitions the shape-annotated computation graph into multiple subgraphs and fine-tunes the backbone operators from the subgraph within a hardware-aligned search space to find a collection of high-performance schedules. TSCompiler can propagate the explored backbone schedule to other fusion groups within the same subgraph to generate a set of parameterized tensor programs for fused cases based on dependence analysis. At runtime, TSCompiler utilizes an occupancy-targeted cost model to select from pre-compiled tensor programs for varied tensor shapes. Extensive evaluations show that TSCompiler can achieve state-of-the-art speedups for dynamic shape models. For example, we can improve kernel efficiency by up to 3.97× on NVIDIA RTX3090, and 10.30× on NVIDIA A100 and achieve up to five orders of magnitude speedups on end-to-end latency.
Conductive adhesive bonding technology is widely used in the lightweight microwave components of satellite payloads. In the space environment, the conductive adhesive is subjected to the coupled action of irradiation ...
详细信息
Optical spectrum analysis provides a wealth of information about the physical *** the development of optical spectrum analysis,sensitivity has been one of the major topics and has become essential in applications deal...
详细信息
Optical spectrum analysis provides a wealth of information about the physical *** the development of optical spectrum analysis,sensitivity has been one of the major topics and has become essential in applications dealing with faint *** high-sensitivity optical detection technologies have been applied in optical spectrum analysis to enhance its sensitivity to single-photon *** an emerging single-photon detection technology,superconducting nanowire single-photon detectors(SNSPDs)have many impressive features such as high detection efficiency,broad operation bandwidth,small timing jitter,and so on,which make them promising for enhancing the performance of optical spectral *** schemes for photon-counting spectrometers based on SNSPDs have been *** article reviews these impressive works and prospects for the future development of this *** breakthroughs can be expected in its theories,device performance,applications,and combinations with in-sensor computing,promoting it to be a mature and versatile solution for optical spectrum analysis on ultra-faint light.
A dual-arm nursing robot can gently lift patients and transfer them between a bed and a *** its lightweight design,high load-bearing capacity,and smooth surface,the coupled-drive joint is particularly well suited for ...
详细信息
A dual-arm nursing robot can gently lift patients and transfer them between a bed and a *** its lightweight design,high load-bearing capacity,and smooth surface,the coupled-drive joint is particularly well suited for these ***,the coupled nature of the joint disrupts the direct linear relationship between the input and output torques,posing challenges for dynamic modeling and practical *** study investigated the transmission mechanism of this joint and employed the Lagrangian method to construct a dynamic model of its internal *** on this foundation,the Newton-Euler method was used to develop a dynamic model for the entire robotic arm.A continuously differentiable friction model was incorporated to reduce the vibrations caused by speed transitions to *** experimental method was designed to compensate for gravity,inertia,and modeling errors to identify the parameters of the friction *** method establishes a mapping relationship between the friction force and motor *** addition,a Fourier series-based excitation trajectory was developed to facilitate the identification of the dynamic model parameters of the robotic *** tracking experiments were conducted during the experimental validation phase,demonstrating the high accuracy of the dynamic model and the parameter identification method for the robotic *** study presents a dynamic modeling and parameter identification method for coupled-drive joint robotic arms,thereby establishing a foundation for motion control in humanoid nursing robots.
Live video streaming has become an important form of communication such as virtual conferences. However, for cross-language communication in live video streaming, reading subtitles degrades the viewing experience. To ...
详细信息
In the process of metal processing, heat treatment is a common metal processing method, which is usually used to change the mechanical properties of metal alloys to control such as hardness, strength, toughness, ducti...
详细信息
The potentials of rare earth-based nanocomposite alloys have never been realized due to strict microstructural *** to the easy demagnetization it is challenging to increase the soft magnetic phase *** avoid the easy d...
详细信息
The potentials of rare earth-based nanocomposite alloys have never been realized due to strict microstructural *** to the easy demagnetization it is challenging to increase the soft magnetic phase *** avoid the easy demagnetization,Pr-Fe-B/Alnico magnets were fabricated and reported in this *** content of the Alnico phase is increased from 0 to 25 wt%,while the content of Pr element is reduced to below the sub-stoichiometry of the 2:14:1 main *** maximum magnetic energy product,which is the figure-of-merit for permanent magnets,is increased from 122 kJ/m^(3) for the standard alloy to 146 kJ/m^(3) for the alloy with 15 wt% Alnico which shows a significant improvement considering the fact that the Curie point of the magnet is also increased by~66 *** special microstructure contains distinctly and heterogeneously distributed 2:14:1 and Alnico *** dimensions of neither the 2:14:1 nor the Alnico phases meet the dimensional requirements of the nanocomposite magnets,but still the smooth demagnetization curves are noted for the *** behavior of effective anisotropy,the performance of the magnets in applied magnetic field and the magnetic interactions among the various constituent grains were quantitatively studied by reversible susceptibility,irreversible susceptibility and re coil loop *** study may provide some guiding principles for the development of nanocomposite magnetic alloys with excellent magnetic properties by using much less RE elements.
Smarter learning plays an important role in promoting the reform of higher education and teaching. This paper first analyzes the problems of intelligent teaching in the field of medical education. On the basis of the ...
详细信息
Scalable,high-capacity,and low-power computing architecture is the primary assurance for increasingly manifold and large-scale machine learning *** electronic artificial agents by conventional power-hungry processors ...
详细信息
Scalable,high-capacity,and low-power computing architecture is the primary assurance for increasingly manifold and large-scale machine learning *** electronic artificial agents by conventional power-hungry processors have faced the issues of energy and scaling walls,hindering them from the sustainable performance improvement and iterative multi-task *** to another modality of light,photonic computing has been progressively applied in high-efficient neuromorphic ***,we innovate a reconfigurable lifelong-learning optical neural network(L2 ONN),for highly-integrated tens-of-task machine intelligence with elaborated algorithm-hardware *** from the inherent sparsity and parallelism in massive photonic connections,L2 ONN learns each single task by adaptively activating sparse photonic neuron connections in the coherent light field,while incrementally acquiring expertise on various tasks by gradually enlarging the *** multi-task optical features are parallelly processed by multi-spectrum representations allocated with different *** evaluations on freespace and on-chip architectures confirm that for the first time,L2 ONN avoided the catastrophic forgetting issue of photonic computing,owning versatile skills on challenging tens-of-tasks(vision classification,voice recognition,medical diagnosis,etc.)with a single ***,L2 ONN achieves more than an order of magnitude higher efficiency than the representative electronic artificial neural networks,and 14×larger capacity than existing optical neural networks while maintaining competitive performance on each individual *** proposed photonic neuromorphic architecture points out a new form of lifelong learning scheme,permitting terminal/edge AI systems with light-speed efficiency and unprecedented scalability.
暂无评论