Based on the effect hierarchy principle in experimental design, an aliased effect-number pattern (AENP, or AP for short) is proposed to judge two-level regular designs; it contains the basic ...information of all effects aliased with other effects at varying severity degrees in a design. Based on the AENP, a general minimum lower-order confounding (GMLOC, or GMC for short) criterion is proposed, and several results follow. First, the word-length pattern, as the core of the minimum aberration (MA) criterion, is a function of the AENP. The same also holds for the clear effects (CE) criterion. Furthermore, the estimation capacity (EC) of a design can be also calculated as a function of the new pattern, and links between the MA and CE criteria are discovered. In addition, a concept of estimation ability is introduced, and it is concluded that a GMC design is the one with the best estimation ability. Finally, more applications of the new pattern are given. All GMC designs of 16 and 32 runs, a number of GMC designs of 64 runs, and some comparisons with the optimal designs under MA and CE criteria are tabulated.
Background
The exact animal origin of severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) remains obscure and understanding its host range is vital for preventing interspecies transmission.
...Methods
Herein, we applied single‐cell sequencing to multiple tissues of 20 species (30 data sets) and integrated them with public resources (45 data sets covering 26 species) to expand the virus receptor distribution investigation. While the binding affinity between virus and receptor is essential for viral infectivity, understanding the receptor distribution could predict the permissive organs and tissues when infection occurs.
Results
Based on the transcriptomic data, the expression profiles of receptor or associated entry factors for viruses capable of causing respiratory, blood, and brain diseases were described in detail. Conserved cellular connectomes and regulomes were also identified, revealing fundamental cell‐cell and gene‐gene cross‐talks from reptiles to humans.
Conclusions
Overall, our study provides a resource of the single‐cell atlas of the animal kingdom which could help to identify the potential host range and tissue tropism of viruses and reveal the host‐virus co‐evolution.
Single‐cell atlases of organs and PBMCs in 20 species were constructed. The expression patterns of viral receptors from 24 virus families were characterised in different tissues or PBMC samples to generate a conclusive view on the cellular tropism of viruses. An exploration of cellular communication in brain cells within 11 species was conducted.
When designing an experiment, it is important to choose a design that is optimal under model uncertainty. The general minimum lower-order confounding (GMC) criterion can be used to control aliasing ...among lower-order factorial effects. A characterization of GMC via complementary sets was considered in Zhang and Mukerjee (2009a); however, the problem of constructing GMC designs is only partially solved. We provide a solution for two-level factorial designs with n factors and N = 2n−m runs subject to a restriction on (n,N): 5N/16 + 1 ≤ n ≤ N − 1. The construction is quite simple: every GMC design, up to isomorphism, consists of the last n columns of the saturated 2(N−1)−(N−1−n+m) design with Yates order. In addition, we prove that GMC designs differ from minimum aberration designs when (n, N) satisfies either of the following conditions: (i) 5N/16 + 1 ≤ n ≤ N/2 − 4, or (ii) n ≥ N/2, 4 ≤ n + 2r − N ≤ 2r−1 − 4 with r ≥ 4.
Real-time video analytics systems typically place models with fewer weights on edge devices to reduce latency. The distribution of video content features may change over time for various reasons ...(i.e. light and weather change) , leading to accuracy degradation of existing models, to solve this problem, recent work proposes a framework that uses a remote server to continually train and adapt the lightweight model at edge with the help of complex model. However, existing analytics approaches leave two challenges untouched: firstly, retraining task is compute-intensive, resulting in large model update delays; secondly, new model may not fit well enough with the data distribution of the current video stream. To address these challenges, in this paper, we present EdgeSync, EdgeSync filters the samples by considering both timeliness and inference results to make training samples more relevant to the current video content as well as reduce the update delay, to improve the quality of training, EdgeSync also designs a training management module that can efficiently adjusts the model training time and training order on the runtime. By evaluating real datasets with complex scenes, our method improves about 3.4% compared to existing methods and about 10% compared to traditional means.
Clear effects criterion is an important criterion for selecting fractional factorial designs 1. Tang et al. 2 derived upper and lower bounds on the maximum number of clear two-factor interactions ...(2fi's) in
2
n-(n-k)
designs of resolution III and IV by constructing
2
n-(n-k)
designs. But the method in 2 does not perform well sometimes when the resolution is III. This article modifies the construction method for
2
n-(n-k)
designs of resolution III in 2. The modified method is a great improvement on that used in 2.
O1; Clear effects criterion is an important criterion for selecting fractional factorial designs 1. Tang et al. 2 derived upper and lower bounds on the maximum number of clear two-factor interactions ...(2fi's) in 2n-(n-k) designs of resolution Ⅲ and Ⅳ by constructing 2n-(n-k) designs. But the method in 2 does not perform well sometimes when the resolution is III. This article modifies the construction method for 2n-(n-k) designs of resolution Ⅲ in 2. The modified method is a great improvement on that used in 2.
Despite the advancements of open-source large language models (LLMs), e.g., LLaMA, they remain significantly limited in tool-use capabilities, i.e., using external tools (APIs) to fulfill human ...instructions. The reason is that current instruction tuning largely focuses on basic language tasks but ignores the tool-use domain. This is in contrast to the excellent tool-use capabilities of state-of-the-art (SOTA) closed-source LLMs, e.g., ChatGPT. To bridge this gap, we introduce ToolLLM, a general tool-use framework encompassing data construction, model training, and evaluation. We first present ToolBench, an instruction-tuning dataset for tool use, which is constructed automatically using ChatGPT. Specifically, the construction can be divided into three stages: (i) API collection: we collect 16,464 real-world RESTful APIs spanning 49 categories from RapidAPI Hub; (ii) instruction generation: we prompt ChatGPT to generate diverse instructions involving these APIs, covering both single-tool and multi-tool scenarios; (iii) solution path annotation: we use ChatGPT to search for a valid solution path (chain of API calls) for each instruction. To enhance the reasoning capabilities of LLMs, we develop a novel depth-first search-based decision tree algorithm. It enables LLMs to evaluate multiple reasoning traces and expand the search space. Moreover, to evaluate the tool-use capabilities of LLMs, we develop an automatic evaluator: ToolEval. Based on ToolBench, we fine-tune LLaMA to obtain an LLM ToolLLaMA, and equip it with a neural API retriever to recommend appropriate APIs for each instruction. Experiments show that ToolLLaMA demonstrates a remarkable ability to execute complex instructions and generalize to unseen APIs, and exhibits comparable performance to ChatGPT. Our ToolLLaMA also demonstrates strong zero-shot generalization ability in an out-of-distribution tool-use dataset: APIBench.
With the development of HVDC system and medium voltage DC distribution network, DC transformer is needed for voltage level conversion and power transmission. In this paper, a multi-phase face-to-face ...(F2F) DC transformer topology based on modular multilevel converter (MMC) is proposed to connect two DC systems with different voltage levels and stabilize the lower bus voltage. When the two systems have fixed voltage levels, the topology can increase the transmission power and reduce AC transformer insulation requirements. In this paper, the topology of the proposed DC transformer is introduced, and the control and modulation strategy of the DC transformer are put forward. It can realize large voltage ratio of the system with small ratio AC transformer, and the voltage level conversion of the DC transformer does not depend on the AC transformer. At last, a simulation example is given in MATLAB/Simulink to verify the feasibility of the proposed topology.
Humans possess an extraordinary ability to create and utilize tools, allowing them to overcome physical limitations and explore new frontiers. With the advent of foundation models, AI systems have ...the potential to be equally adept in tool use as humans. This paradigm, i.e., tool learning with foundation models, combines the strengths of specialized tools and foundation models to achieve enhanced accuracy, efficiency, and automation in problem-solving. Despite its immense potential, there is still a lack of a comprehensive understanding of key challenges, opportunities, and future endeavors in this field. To this end, we present a systematic investigation of tool learning in this paper. We first introduce the background of tool learning, including its cognitive origins, the paradigm shift of foundation models, and the complementary roles of tools and models. Then we recapitulate existing tool learning research into tool-augmented and tool-oriented learning. We formulate a general tool learning framework: starting from understanding the user instruction, models should learn to decompose a complex task into several subtasks, dynamically adjust their plan through reasoning, and effectively conquer each sub-task by selecting appropriate tools. We also discuss how to train models for improved tool-use capabilities and facilitate the generalization in tool learning. Considering the lack of a systematic tool learning evaluation in prior works, we experiment with 18 representative tools and show the potential of current foundation models in skillfully utilizing tools. Finally, we discuss several open problems that require further investigation for tool learning. Overall, we hope this paper could inspire future research in integrating tools with foundation models.