Yanhong Huang’s research while affiliated with East China Normal University and other places

... However, as these findings rely on licensed models and cloud solutions, we concentrated our research on models that we can employ locally in resource-restricted and critical environments. Zhao et al. [49] generated a dataset with GPT-3.5 to fine-tune a T5 model, which is used to translate natural language requirements into Computation Tree Logic (CTL). They report that the fine-tuned model achieved an accuracy of 46.4% and therefore led to better results then few-shot learning with ...

Reference:

Overcoming Data Shortage in Critical Domains With Data Augmentation for Natural Language Software Requirements

... These models can flexibly interpret designers' intentions through NL-based instructions, comprehend and reason over the instructions, and generate corresponding modeling commands. Researchers have begun exploring LLM-based tools to support CAD model creation and editing Yuan et al., 2024). However, the potential of LLMs for fast prototyping with its targeted control demands has yet to be fully elicited, particularly in the context of the targetedness-demand fast prototyping (Uusitalo et al., 2024). ...

Reference:

Targeted control of fast prototyping through domain-specific interface

... This increases the need for testing and verification of PLC programs [38], [46]. Existing methods for the automated generation of PLC test cases mainly include symbolic execution [44], concolic testing [7] and search-based techniques [13]. However, these approaches can produce test cases that are difficult to maintain, making them challenging to use. ...

Reference:

Agents4PLC: Automating Closed-loop PLC Code Generation and Verification in Industrial Control Systems using LLM-based Agents

... Formula progression is a common technique that steps through time to evaluate formulas; it dates back to the 1990's [7] and has been developed for various temporal logics including MTL [17], LTL f [41], and MLTL [31]. Formula progression represents a straightforward and generally easy-to-implement algorithm for evaluating whether a particular temporal trace satisfies the given formula, and is therefore utilized in many contexts, including satisfiability solving, validation, runtime verification, and benchmark generation. ...

Reference:

Formalizing MLTL Formula Progression in Isabelle/HOL

... Ma et al. [10] addressed the interpretation of RF based on first-order logic. They encode the decision-making process of these forests into first-order logic formulas, enabling improved explainability of decisions based on the importance of features. ...

Reference:

Formal Analysis of Tree-Based Machine Learning Models Using Coloured Petri Nets

... In the proposed method, the input video would first be pre-processed to extract relevant features. These features could be high-level semantic features obtained from pre-trained models or lower-level features like motion vectors or visual descriptors [76]. ...

Reference:

A SAT-centered XAI method for Deep Learning based Video Understanding

... This approach allows the robot to dock without a complete stop by aligning the direction and distance between the robot and the docking station. Tsiogas [19] introduced a vision-based approach that enables the safe and autonomous operation of pallet moving vehicles, and Ding [20] proposed a combination of the double-wheel chassis model (DCM) and PID control strategy to enhance the precision and stability during docking. Moreover, visual strategies are applied for accurate docking alignment. ...

Reference:

An Accurate and Rapid Docking Algorithm for Four-Way Shuttle in High-Density 3D Warehousing Environment

... In this way, there will be progress problems in expressing the number of floating points. In the flow accumulation, this error will be infinitely amplified, resulting in the inaccuracy of the calculation results [6][7][8][9][10]. This error has a certain impact on production management. ...

Reference:

Research on Floating Point Accumulation Based on Improved Kahan Algorithm

... Due to the fact that these compounds have the capacity to undergo tautomerism, studies of this property have been carried out in the group azo-hydrazone in different solvents in order to examine the affinity of binding with streptavidin by combining spectroscopic methods and computational [16]. This compound can be used for determining the amount and biotin-binding activity of avidin in solution, since avidin is a glycoprotein antagonist of biotin, which binds with high affinity to form an insoluble derivative that prevents its bioavailability [17]. Azobenzene derivatives have been investigated for use in digital data storage [18] as On and off switches of nanomolecular devices and in ion channels [19]. ...

Reference:

Analyses of the structure and supramolecular organisation of 2-E-((4-hydroxyphenyl) diazenyl) benzoic acid

... Xu et al. proposed a formal modeling and verification method to ensure safety of cooperative driving in vehicle platoons. 77 They defined an abstract model to describe the spatial properties of driving scenarios and introduced a timed automata type for constructing platoon decision controller models. By incorporating multi-agent spatial logic (MASL) specifications to define model guard conditions, they implemented the corresponding platoon decision controller model in UPPAAL for a case study involving human-driven vehicles joining the platoon. ...

Reference:

A comprehensive survey of UPPAAL‐assisted formal modeling and verification