The evaluation of classifiers' performances plays a critical role in construction and selection classification model. Although many performance metrics have been proposed machine learning community, no general guidelines are available among practitioners regarding which metric to be selected for evaluating classifier's performance. In this paper, we attempt provide with strategy on selecting classifier evaluation. Firstly, the authors investigate seven widely used metrics, namely accuracy,...

It is known that federated learning (FL) incurs heavy communication overhead for model training by exchanging updates between clients and the parameter server (PS) over Internet multiple rounds. Compressing an effective approach to alleviating in FL. Yet tradeoff compression accuracy networked environment remains unclear and, simplicity, most implementations adopt a fixed rate only during entire process. In this paper, we first time systematically examine tradeoff, explicitly quantifying...

Proof of Authority (PoA) plays a pivotal role in blockchains for reaching consensus. Clique, which selects consensus nodes to generate blocks with pre-determined order, is the most popular implementation PoA due its low communication overhead and energy consumption. However, our study unveils that speed process transactions by Clique severely restricted 1) long delay full (each containing certain number transactions) between nodes; 2) occurrences no-turn blocks, generated if an in-turn block...

This paper reports the results of a study on implementation sustainable teaching model based OBE (Outcome-Based Education) concept and TSEM (Teach, Study, Evaluate, Manage) framework in computer science technology at NingboTech University, China. In context digital education, are integrated to explore learning models “artificial intelligence education”. Based core engineering professional education accreditation, course is designed by using PCCM (Professional Competency Correlation Matrix)...

The increasing demand for precise load forecasting distribution networks has become a crucial requirement due to the continual surge in power consumption. Accurate of peak loads is paramount ensure that grids operate smoothly and optimize their configuration. Many methods do not meet requirements accurate data trend fitting. To address these issues, this paper presents novel model called Prophet-LSTM, which combines strengths Prophet model’s high fitting LSTM prediction accuracy, resulting...

The interactions between various substances in foods have significant implications for their nutritional values. antioxidant capacity of polyphenols tea can also be modulated by with other substances. In this study, the intermolecular hydrogen bonds epigallocatechin gallate (EGCG) and theanine, as well capacity, were both analyzed demonstrated. results showed that antiradical activity against DPPH• increased approximate 2.4 %, relative cellular PC12 cells peroxide 3.4 when appropriate...

There is an emerging trend of integrating machine learning (ML) techniques into database systems (DB). Considering that almost all the ML toolkits assume input algorithms a single table even though many real-world datasets are stored as multiple tables due to normalization in DB. Thus, data scientists have perform joins before model. This strategy called after joins, which incurs redundancy avoided by normalization. In area ML, Support Vector Machine (SVM) one most standard classification...

Reputation is useful for establishing trust between Web service (WS) providers and WS consumers. In the context of composition, a challenging issue reputation management to propagate user's impression composite (i.e., feedback rating) its component WSs. this paper, we propose Shapley value based approach which can achieve fair propagation, that is, never awarded (or penalized) good bad) performances other peer WSs in same WS. The fairness proposed validated through theoretical analysis...

Popular reinforcement learning (RL) algorithms tend to produce a unimodal policy distribution, which weakens the expressiveness of complicated and decays ability exploration. The diffusion probability model is powerful learn multimodal distributions, has shown promising potential applications RL. In this paper, we formally build theoretical foundation representation via provide practical implementations for online model-free Concretely, character as stochastic process, new approach...

High-utility pattern mining is an effective technique that extracts significant information from varied types of databases. However, the analysis data with sensitive private may cause privacy concerns. To achieve better trade-off between utility maximizing and preserving, privacy-preserving (PPUM) has become important research topic in recent years. The MSICF algorithm a sanitization for PPUM. It selects item based on conflict count identifies victim transaction concept utility. Although...

In the cloud service community, providers typically limit component services to perform a specific and limited function maximize versatility of services. However, business needs consumers are often very complex, requiring multiple be combined complete business. process composition, question how find right combination partner is challenging. This paper starts with brief history execution it then analyzes historical information execution. Based on invocation relationship service, community...

The continuing development of intelligent transportation terminals and the massive generated traffic data have placed tremendous pressure on acquisition. However, most existing systems applications rely well-defined data, while few studies focus how to collect live from various vehicle in a large number. To solve this problem, we propose an efficient non-blocking acquisition system paper, which can retrieve based different priority or QoS requirements number real-time, so that application...