Using deep learning models on spatial transcriptomics data to identify the spatial domain is crucial for uncovering the spatial distribution of cells and gene expression patterns within tissues, essential for understa...
详细信息
ISBN:
(数字)9798350386226
ISBN:
(纸本)9798350386233
Using deep learning models on spatial transcriptomics data to identify the spatial domain is crucial for uncovering the spatial distribution of cells and gene expression patterns within tissues, essential for understanding complex biological processes and disease mechanisms. Existing methods for spatial domain partitioning often rely on predefined adjacency relationships at a single scale, overlooking the hierarchical structure and functional characteristics of biological tissues. In this paper, we propose SpaNFM, a novel method that leverages sparse attention-based hierarchical node representation and multi-view contrastive learning for spatial domain identification in spatial transcriptomics data. The SpaNFM first treats each spot as a node and constructs two views using different data augmentation techniques based on tissue image information, gene expression profiles, and spatial coordinates of cells. Subsequently, SpaNFM utilizes a sparse attention-based hierarchical node fusion module to generate coarse-grained node representations. This fine-to-coarse hierarchical structure integrates complementary information from multi-granularity node features and reduces model complexity due to the decreased node size. The model parameters are updated using gene expression reconstruction loss and contrastive loss on the coarse-grained node representations from the two views. Finally, the learned node features are subjected to downstream clustering using the Leiden algorithm. We tested SpaNFM on the human dorsolateral prefrontal cortex dataset. The results demonstrate that SpaNFM outperforms other state-of-the-art methods in most cases. The data and code are available at: https://***/weiba/SpaNFM
Visual recognition is currently one of the most important and active research areas in computer vision,pattern recognition,and even the general field of artificial *** has great fundamental importance and strong indus...
详细信息
Visual recognition is currently one of the most important and active research areas in computer vision,pattern recognition,and even the general field of artificial *** has great fundamental importance and strong industrial needs,particularly the modern deep neural networks(DNNs)and some brain-inspired methodologies,have largely boosted the recognition performance on many concrete tasks,with the help of large amounts of training data and new powerful computation *** recognition accuracy is usually the first concern for new progresses,efficiency is actually rather important and sometimes critical for both academic research and industrial ***,insightful views on the opportunities and challenges of efficiency are also highly required for the entire *** general surveys on the efficiency issue have been done from various perspectives,as far as we are aware,scarcely any of them focused on visual recognition systematically,and thus it is unclear which progresses are applicable to it and what else should be *** this survey,we present the review of recent advances with our suggestions on the new possible directions towards improving the efficiency of DNN-related and brain-inspired visual recognition approaches,including efficient network compression and dynamic brain-inspired *** investigate not only from the model but also from the data point of view(which is not the case in existing surveys)and focus on four typical data types(images,video,points,and events).This survey attempts to provide a systematic summary via a comprehensive survey that can serve as a valuable reference and inspire both researchers and practitioners working on visual recognition problems.
Modern diagnostic systems for asynchronous motors are widely used in the power industry. The most promising diagnostic systems for rotor eccentricity of induction motors are systems with capacitive measuring transduce...
详细信息
A two-layer labeling-based dependency analysis model is proposed to introduce the location features of words for the coreference disambiguation of entities. Firstly, two layers of labels are used for labeling, the sec...
详细信息
In the dynamic landscape of energy efficiency, a domain garnering ever increasing attention and urgency, the cornerstone of informed decision-making resides in the acquisition of reliable data. This paper summarizes t...
详细信息
ISBN:
(数字)9798350379730
ISBN:
(纸本)9798350379747
In the dynamic landscape of energy efficiency, a domain garnering ever increasing attention and urgency, the cornerstone of informed decision-making resides in the acquisition of reliable data. This paper summarizes the main approaches, methods and hardware used to gather, process and disaggregate such data on an appliance-by-appliance basis. The paper explores a critical analysis of the various models and techniques that support the gathering of energy data at the appliance level. A range of approaches, including hardware-based solutions such as smart meters, submeters, and sensor arrays, placing a special focus on intrusive load monitoring (ILM) and non-intrusive load monitoring (NILM) algorithms that employ advanced signal processing and machine learning techniques, are examined for their effectiveness, scalability, and suitability for both industrial and commercial contexts. Furthermore, the challenges of data acquisition capabilities, reconciling scalability, cost-effectiveness, and user-friendliness - prerequisites for widespread acceptance and uptake are also investigated.
Farmers in rural areas often struggle to access crucial agricultural information due to language barriers, low literacy rates, and limited exposure to digital tools. While many can write in Bangla, most agricultural r...
详细信息
Fault diagnosis in wastewater treatment plants (WWTPs) is important to protect communities and ecosystems from toxic elements discharged into water. In this sense, fault identification of sensors plays an important ro...
Fault diagnosis in wastewater treatment plants (WWTPs) is important to protect communities and ecosystems from toxic elements discharged into water. In this sense, fault identification of sensors plays an important role as they are the key components of the water plants control, especially because environmental legislation is very strict when referring to failures or anomalies in WWTPs. This paper analyzes the performances of two Deep Learning models, a Feedforward Neural Network (FFNN) and a 1D Convolution Neural Network (1DCNN) for identifying five operating states of the dissolved oxygen (DO) sensor: normal and faulty (bias, stuck, spike and precision degradation faults). The experiments were conducted on the Benchmark Simulator Model No 2 (BSM2) developed by the IWA Task Group. The performance of the Deep Learning (DL) classifiers was evaluated via accuracy, precision, recall, and F1-score metrics. The best overall classification accuracy was obtained by FFNN, 98.32% for training and 98.30% for testing.
The skin lesion can be thought of as a biological system, so the morpho-granulometry of significant color clusters found in skin lesions is one of the elements that reproduce in a natural way the structure of the lesi...
详细信息
ISBN:
(数字)9798350364293
ISBN:
(纸本)9798350364309
The skin lesion can be thought of as a biological system, so the morpho-granulometry of significant color clusters found in skin lesions is one of the elements that reproduce in a natural way the structure of the lesion, this novelty is highlighted in this study. Important features of skin lesions can be modulated by fusing neural networks (NN) and machine learning (ML). By choosing the nevus and melanoma classes, the primary goal was accomplished, and three databases were used to test the methodology. The characteristics based on morpho-granulometry allowed for the identification of microstructure within the images, which can be very helpful in characterizing the biological system. Based on random forest (RF) and extreme gradient boosting (XGboost) classifiers, this work aimed to improve the classification performance of important feature selection. The selected features from three free image databases with three NNs were classified. In a binary classification of nevus vs. melanoma, the results showed that the pattern recognition neural network (PRNN), according to the PH2 database, provided an accuracy of 0.923 and an F1-score of 0.876. The classification is interpretable if it is not validated. In our study, the best results were verified with a logistic regression (LR) classifier.
This paper focuses on analyzing power quality issues resulting from electric arc furnace operation. It highlights the challenges posed by the stochastic nature of the arc and proposes strategies to mitigate its impact...
详细信息
ISBN:
(数字)9798350355185
ISBN:
(纸本)9798350355192
This paper focuses on analyzing power quality issues resulting from electric arc furnace operation. It highlights the challenges posed by the stochastic nature of the arc and proposes strategies to mitigate its impact on the power grid. Through measurements carried out on the medium and low voltage side of the furnace, fluctuations in parameters such as voltage, current, and power factor were observed, indicating significant disturbances in power quality. Mitigation strategies, including e.g. installing compensators, are suggested to stabilize the network and improve power quality, enhancing efficiency and reducing energy consumption and electrode usage.
Unlike the case with identical neighboring agents whose actions are mirrored, the problem of distributed formation control design with heterogeneous sensing is not straightforward. In this paper, we consider the probl...
详细信息
暂无评论