While Math Word Problem (MWP) solving has emerged as a popular field of study and made great progress in recent years, most existing methods are benchmarked solely on one or two datasets implemented with different configurations. In this paper, we introduce the first open-source library for MWPs called MWPToolkit, which provides unified, comprehensive, extensible framework research purpose. Specifically, deploy 17 deep learning-based MWP solvers 6 our toolkit. These advanced models solving,...

Current constrained reinforcement learning (RL) methods guarantee constraint satisfaction only in expectation, which is inadequate for safety-critical decision problems. Since a satisfied expectation remains high probability of exceeding the cost threshold, solving RL problems with probabilities critical safety. In this work, we consider safety criterion as on conditional value-at-risk (CVaR) cumulative costs, and propose CVaR-constrained policy optimization algorithm (CVaR-CPO) to maximize...

Developing automatic Math Word Problem (MWP) solvers has been an interest of NLP researchers since the 1960s. Over last few years, there are a growing number datasets and deep learning-based methods proposed for effectively solving MWPs. However, most existing benchmarked soly on one or two datasets, varying in different configurations, which leads to lack unified, standardized, fair, comprehensive comparison between methods. This paper presents MWPToolkit, first open-source framework In we...

Abstract Background Inhibiting secondary inflammatory damage caused by glial cells and creating a stable microenvironment is one of the main strategies to investigate drugs for treatment spinal cord injury. Acetyl‐11‐keto‐beta‐boswellic acid (AKBA) active component natural drug boswellia, which has anti‐inflammatory antioxidant effects offers possible therapeutic option Methods In this study, injury model was established crushing cord, respectively, detect M1 macrophage markers: iNOS, TNF‐α,...

While diverse question answering (QA) datasets have been proposed and contributed significantly to the development of deep learning models for QA tasks, existing fall short in two aspects. First, we lack covering complex questions that involve answers as well reasoning processes get them. As a result, state-of-the-art research on numerical still focuses simple calculations does not provide mathematical expressions or evidence justifying answers. Second, community has lot effort improve...

Lodging is a key factor affecting maize yield and harvestability. This study utilized Reid population baselines their improved lines as female parents No-Reid male to form 48 incomplete diallel crosses. The genetic improvement effects, combining ability, heterosis of three lodging resistance-related traits (stem tension, puncture strength, crushing strength at the third internode) were analyzed. Regarding improvement, results indicated that all significantly in compared baselines, with...

The primary aim of this study is to delve into the potential Acetyl-11-keto-β-boswellic acid (AKBA) in ameliorating neuronal damage induced by acute spinal cord injury, as well unravel intricate underlying mechanisms. A cohort 40 Sprague-Dawley rats was meticulously categorized four groups. Following a seven-day oral administration AKBA, damaged samples were procured for Nissl staining and electron microscopy assess demise. Employing ELISA, immunofluorescence, Western blot (WB), quantitative...

The existing studies by our team demonstrated the pro-recovery effect of 3-Acetyl-11-keto-beta-boswellic acid (AKBA) on a sciatic nerve injury. To further investigate role AKBA in peripheral injury repair, TMT quantitative proteomics technique was used to obtain differentially significant proteins Sham group, Model and group. After that, three time points (5, 14, 28 d) four groups (Sham + AKBA, Sham, Model, AKBA) were set up, immunoblotting, immunofluorescence, cellular assays applied...

Authentication is necessary to not only data exchange processes but also network administrative tasks in wireless sensor networks. So far, authentication has drawn much attention from the research community. We propose an efficient distributed node (DNA) scheme that utilizes geographic location and trust relationship among neighboring nodes authenticate identity of nodes. Analysis simulation results demonstrate DNA robust against failure/sleeping resilient attackers.

This work focuses on motion control of the bicycle robot by using proposed NRRL algorithm. Unlike traditional RL algorithm, decomposing main tasks into subtasks manually and introducing qualitative prior knowledge to agent have been applied in Simulation results show that better performance sample efficiency algorithm achieved terms balance path tracking robot. It's believed is available real physical robot, deployment will be realized soon, as has constructed currently.

In this paper, we propose an intrusion detection framework for smartphone systems. We formulate the problem into a two-player, non-cooperative, complete-information, constant-sum game. The attacker and security server are players of wants to maximize value system but minimize it. present Nash equilibrium leads defense strategy server. implement results show that proposed is better than traditional ones.

Reinforcement Learning (RL) agents are often fed with large-dimensional observations to achieve the ideal performance in complex environments. Unfortunately, massive observation space usually contains useless or even adverse features, which leads low sample efficiency. Existing methods rely on domain knowledge and cross-validation discover efficient features informative for decision-making. To minimize impact of prior knowledge, we propose a temporal-adaptive feature attention algorithm...

Unsupervised pre-training in reinforcement learning enables the agent to gain prior environmental knowledge, which is then fine-tuned supervised stage quickly adapt various downstream tasks. In absence of task-related rewards, aims acquire policies (i.e., behaviors) that generate different trajectories explore and master environment. Previous research categorizes states into their associated behaviors by a discriminator. However, an underlying problem persists: such discriminator trained...

Comprehensively understanding and accurately predicting the performance of large language models across diverse downstream tasks has emerged as a pivotal challenge in NLP research. The pioneering scaling law on works demonstrated intrinsic similarities within model families utilized such for prediction. However, they tend to overlook between only consider design factors listed original law. To overcome these limitations, we introduce novel framework, Collaborative Performance Prediction...

This paper focuses on the problem of protocol identification in industrial internet and proposes an unknown method. We first establish detection model to classify known protocols, interference signals then store protocols for manual analysis. Based Eps-neighborhood idea, we further develop hit algorithm propose method identify where supervised learning classification is realized. Finally, extensive experimental results are provided illustrate our theoretical findings. It indicates that...

Deep reinforcement learning (DRL) has been proved to be a very promising method for robot motion control. However, it usually needs large number of samples in training, which restricts its application real-world robots. Sim-to-Real means transferring training strategies simulation reality, and become one the hottest research areas recent years. Embodiment DRL algorithms is necessary step Sim-to-Real, faced with many challenges, such as poor sample efficiency, wear tear robots, safety, etc....

Many real-world applications require intelligent agents to coordinate with other in complex environments. As a stepping stone this goal, the domain of multi-agent cooperative search has emerged as an important challenge for swarm intelligence research, owing its endogenous complicated behavior mechanism and interaction uncertain dynamic environment. We choose address by proposing decentralized reinforcement learning method DELTAS team unmanned aerial vehicles perform efficient area interest....

In recent years, the objectification of tongue image in Traditional Chinese Medicine (TCM) has become a topic extensive discussion. As an important reference index TCM diagnosis, features play role disease prevention and diagnosis. Among them, how to achieve standardization accurate segmentation extracting quantitative feature information key issue that restricts objective development This article mainly reviews two aspects color processing methods. Review currently widely used methods,...

Boswellia is a traditional medicine for bruises and injuries. Its main active ingredient, acetyl-11-keto-beta-boswellic acid, has antioxidant antiapoptotic effects. In this experiment, we used Sprague-Dawley rats to make sciatic nerve injury model detect the transcription factor NF-E2-related 2/heme oxygenase 1 signaling pathway apoptosis, combined with clinical indicators, testing whether acid can reduce oxidative stress promote repair. Our results showed that administration promoted myelin...

The non-standard machinery refers to customized produced meet specific customer demands. mainstream research direction in data stream anomaly detection focuses on deep learning, which involves learning distribution through a large amount of training data. However, equipment has the characteristics small production scale and sparse samples, making it difficult obtain sufficient annotated sets. This inadequacy results model not enough, thereby rendering unable effectively detect abnormal...