In recent years, machine learning techniques have been widely used to solve many problems for fault diagnosis. However, in real-world diagnosis applications, the distribution of source domain data (on which model is trained) different from target (where learned actually deployed), leads performance degradation. this paper, we introduce adaptation, can find solution problem by adapting classifier or regression trained a use but related domain. particular, proposed novel deep neural network...

The Iterative Closest Points (ICP) algorithm is the mainstream used in process of accurate registration 3D point cloud data. requires a proper initial value and approximate two clouds to prevent from falling into local extremes, but actual matching process, it difficult ensure compliance with this requirement. In paper, we proposed ICP based on features (GF-ICP). This method uses geometrical be registered, such as curvature, surface normal density, search for correspondence relationships...

In recent years, intelligent fault diagnosis technology with the deep learning algorithm has been widely used in manufacturing industry for substituting time-consuming human analysis method to enhance efficiency of diagnosis. The rolling bearing as connection between rotor and support is crucial component rotating equipment. However, working condition under changing complex operation demand, which will significantly degrade performance method. this paper, a new transfer model based on...

As bearings are the most common components of mechanical structure, it will be helpful to research bearing fault and diagnose as early possible in case suffering greater losses. This paper proposes a deep neural network algorithm framework for diagnosis based on stacked autoencoder softmax regression. The simulation results verify feasibility show excellent classification performance. In addition, this represents strong robustness eliminates impact noise remarkably. Last but not least, an...

Time-of-Flight (ToF) cameras, a technology which has developed rapidly in recent years, are 3D imaging sensors providing depth image as well an amplitude with high frame rate. As ToF camera is limited by the conditions and external environment, its captured data always subject to certain errors. This paper analyzes influence of typical distractions including material, color, distance, lighting, etc. on error cameras. Our experiments indicated that factors such distance could cause different...

Abstract With the breakthrough of AlphaGo, human-computer gaming AI has ushered in a big explosion, attracting more and researchers all over world. As recognized standard for testing artificial intelligence, various systems (AIs) have been developed, such as Libratus, OpenAI Five, AlphaStar, which beat professional human players. The rapid development AIs indicates step decision-making it seems that current techniques can handle very complex games. So, one natural question arises: What are...

Abstract With the breakthrough of AlphaGo, deep reinforcement learning has become a recognized technique for solving sequential decision-making problems. Despite its reputation, data inefficiency caused by trial and error mechanism makes difficult to apply in wide range areas. Many methods have been developed sample efficient learning, such as environment modelling, experience transfer, distributed modifications, among which shown potential various applications, human-computer gaming...

Recently, deep multi-agent reinforcement learning (MARL) has shown the promise to solve complex cooperative tasks. Its success is partly because of parameter sharing among agents. However, such may lead agents behave similarly and limit their coordination capacity. In this paper, we aim introduce diversity in both optimization representation shared learning. Specifically, propose an information-theoretical regularization maximize mutual information between agents' identities trajectories,...

Widely used in three-dimensional (3D) modeling, reverse engineering and other fields, point cloud registration aims to find the translation rotation matrix between two clouds obtained from different perspectives, thus correctly match clouds. As most common method, ICP algorithm, however, requires a good initial value, not too large transformation clouds, also much occlusion; Otherwise, iteration would fall into local minimum. To solve this problem, paper proposes an algorithm based on...

Feature extraction plays a significant role in the rolling bearing fault diagnosis. However, complexity and variability of actual working condition leads to data unstable characteristics unpredictable. Traditional machine learning methods won't work or have poor performance under this circumstance. In paper, we propose cross domain feature method based on Transfer Component Analysis algorithm solve problem. Analysis, as novel transfer field, is an efficient for The proposed verified with...

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...

It is a critical issue that whether or not the extremely deep solar minimum of cycle 23/24 brought serious influences on Earth's space environment. In this study, we collected and manually scaled ionograms recorded by DPS ionosonde at Jicamarca (12.0°S, 283.2°E) to retrieve F layer parameters electron density ( N e ) profiles. A comparative study performed evaluate equatorial ionosphere in minima 22/23 (1996–1997) (2008–2009). The seasonal median values frequency 2 f o were remarkably...

As a key on-orbit service technology, relative measurement of non-cooperative spatial targets can give accurate pose for the unmanned rendezvous and docking targets. With rapid development three-dimensional data acquisition equipment such as flash lidar in field targets' recent years, research on tracking technology based (3D) point cloud has become more urgent. In this paper, we proposed an approach autonomous recognition features. For method, density, curvature normal angle target 3D are...

Obstructive sleep apnea (OSA) is regarded as one of the most common sleep-related breathing disorders, which causes various diseases and affects people's daily life severely. Up to now, massive efforts have been devoted identifying OSA events during based on different signals (e.g., PSG, ECG, nasal airflow EMG, etc.). However, there still are more or less shortcomings in current studies. In this paper, we propose a novel framework improve performance events. Particularly, key idea our divide...

Among the remarkable successes of Reinforcement Learning (RL), self-play algorithms have played a crucial role in solving competitive games. However, current RL methods commonly optimize agent to maximize expected win-rates against its or historical copies, resulting limited strategy style and tendency get stuck local optima. To address this limitation, it is important improve diversity policies, allowing break stalemates enhance robustness when facing with different opponents. In paper, we...

Rolling bearing devices are widely used in almost all industries the world, and play a very critical role. So that once this device fails, whole system will have serious impact. It not only affect performance of entire system, but also reliability, security, applicability so on. Therefore, it is important to predict failure. Because recurrent neural network quite effective dealing with sequence problems, often do prediction-related problems. And recent years, has been put into great...

The star centroid estimation is the most important operation, which directly affects precision of attitude determination for sensors. This paper presents a theoretical study systematic error introduced by algorithm. analyzed through frequency domain approach and numerical simulations. It shown that consists approximation truncation resulted from discretization sampling window limitations, respectively. A criterion choosing size to reduce given in this paper. can be evaluated as function...

On-line decision augmentation (OLDA) has been considered as a promising paradigm for real-time making powered by Artificial Intelligence (AI). OLDA widely used in many applications such fraud detection, personalized recommendation, etc. inference puts features extracted from multiple time windows through pre-trained model to evaluate new data support making. Feature extraction is usually the most time-consuming operation pipelines. In this work, we started studying how existing in-memory...