Model Expression Research Articles

A Multi-Feature Fusion Method for Life Prediction of Automotive Proton Exchange Membrane Fuel Cell Based on TCN-GRU.

The Proton Exchange Membrane Fuel Cell (PEMFC) is a fast-developing battery technology, and the key to its reliability and lifespan improvement lies in the accurate assessment of durability. However, the degradation mechanism of the PEMFC is hard to determine and its internal parameters are highly coupled. Thus, the development of a more accurate life prediction model that meets the actual scenarios is to be investigated urgently. To solve this problem, a multi-feature fusion life prediction method based on the Temporal Convolutional Network-Gated Recurrent Unit (TCN-GRU) is proposed. A TCN algorithm is used as the prediction base model, and two GRU modules are included with the model to strengthen the model's expression ability and improve its predictive accuracy. Two widely recognized datasets and two operating conditions are utilized for model training and prediction, respectively. Comparisons are made with single-feature parameter models in terms of Root Mean Square Error (RMSE) and the Determination Coefficient (R2). The results show that the prediction accuracy of the TCN-GRU multi-feature fusion model is higher than that of the single-feature models in terms of stability and anti-interference under both operating conditions. The accuracy of the TCN-GRU (three-feature) model is the most optimal in a steady-state condition at 80% of the training set ratio (RMSE = 3.27 × 10-3, R2 = 0.965). Furthermore, with the increase in the input feature parameter, the TCN-GRU model is closer to the real value, which proves once again that the proposed model can meet the accuracy requirements of the life prediction of the PEMFC.

Enhancing Major Depressive Disorder Diagnosis With Dynamic-Static Fusion Graph Neural Networks.

Major Depressive Disorder (MDD) is a debilitating, complex mental condition with unclear mechanisms hindering diagnostic progress. Research links MDD to abnormal brain connectivity using functional magnetic resonance imaging (fMRI). Yet, existing fMRI-based MDD models suffer from limitations, including neglecting dynamic network traits, lacking interpretability, and struggling with small datasets. We present DSFGNN, a novel graph neural network framework addressing these issues for improved MDD diagnosis. DSFGNN employs a graph isomorphism encoder to model static and dynamic brain networks, achieving effective fusion of temporal and spatial information through a spatiotemporal attention mechanism, thereby enhancing interpretability. Furthermore, we incorporate a causal disentangling module and orthogonal regularization module to augment the model's expressiveness. We evaluate DSFGNN on the Rest-meta-MDD dataset, yielding superior results compared to the best baseline. Besides, extensive ablation studies and interpretability analysis confirm DSFGNN's effectiveness and potential for biomarker discovery.

Model of Arteriovenous Malformation Created in Human Placenta for Training in Vascular Microneurosurgery.

Arteriovenous malformations (AVMs) are congenital lesions, and because of their structure, complexity, flow, size, and location organization, they are lesions that require extensive anatomic knowledge and mastery of microsurgical skills and techniques. Human placentas as a training model for AVM surgery are promising alternatives. This article aims to describe the technique for forming an AVM-type lesion in human placentas and its usefulness in the training of microsurgical treatment techniques. In this study, 15 fresh human placental models were treated. A nidus was created using synthetic material, and dynamic flow was evaluated with intravascular injection of Indocyanine Green. The catheter system was connected to a continuous flow infusion pump. For simulation purposes, 4 vascular neurosurgeons and 4 vascular neurosurgery fellows used the same techniques and instruments used in real surgery to simulate the resection of AVM lesions. Subjective assessments were conducted, evaluating the validity and structured content on a 5-point Likert scale. Evaluation criteria included the execution of technical maneuvers and the model's expression and structural aspects. We describe the step-by-step creation of an AVM in a placental biological model for the performance of vascular microsurgery training in the laboratory. We created in the human placenta a lesion with the characteristics of an AVM for microsurgical training in the laboratory, which presents key features realistic to a real AVM, such as 1 or more feeder arteries, nidus (synthetic), draining vein(s), continuous and pulsatile flow, and 3-dimensional configuration. Furthermore, it demonstrates the applicability of microsurgical techniques to the model compared with performing surgery on a patient. Considering it an effective method for laboratory training, the creation of arteriovenous malformations in human placentas enables students to replicate, comprehend the structure, and master microsurgical techniques in a realistic model.

A lightweight Yunnan Xiaomila detection and pose estimation based on improved YOLOv8.

Yunnan Xiaomila is a pepper variety whose flowers and fruits become mature at the same time and multiple times a year. The distinction between the fruits and the background is low and the background is complex. The targets are small and difficult to identify. This paper aims at the problem of target detection of Yunnan Xiaomila under complex background environment, in order to reduce the impact caused by the small color gradient changes between xiaomila and background and the unclear feature information, an improved PAE-YOLO model is proposed, which combines the EMA attention mechanism and DCNv3 deformable convolution is integrated into the YOLOv8 model, which improves the model's feature extraction capability and inference speed for Xiaomila in complex environments, and achieves a lightweight model. First, the EMA attention mechanism is combined with the C2f module in the YOLOv8 network. The C2f module can well extract local features from the input image, and the EMA attention mechanism can control the global relationship. The two complement each other, thereby enhancing the model's expression ability; Meanwhile, in the backbone network and head network, the DCNv3 convolution module is introduced, which can adaptively adjust the sampling position according to the input feature map, contributing to stronger feature capture capabilities for targets of different scales and a lightweight network. It also uses a depth camera to estimate the posture of Xiaomila, while analyzing and optimizing different occlusion situations. The effectiveness of the proposed method was verified through ablation experiments, model comparison experiments and attitude estimation experiments. The experimental results indicated that the model obtained an average mean accuracy (mAP) of 88.8%, which was 1.3% higher than that of the original model. Its F1 score reached 83.2, and the GFLOPs and model sizes were 7.6G and 5.7MB respectively. The F1 score ranked the best among several networks, with the model weight and gigabit floating-point operations per second (GFLOPs) being the smallest, which are 6.2% and 8.1% lower than the original model. The loss value was the lowest during training, and the convergence speed was the fastest. Meanwhile, the attitude estimation results of 102 targets showed that the orientation was correctly estimated exceed 85% of the cases, and the average error angle was 15.91°. In the occlusion condition, 86.3% of the attitude estimation error angles were less than 40°, and the average error angle was 23.19°. The results show that the improved detection model can accurately identify Xiaomila targets fruits, has higher model accuracy, less computational complexity, and can better estimate the target posture.

Advancing cancer driver gene detection via Schur complement graph augmentation and independent subspace feature extraction

Accurately identifying cancer driver genes (CDGs) is crucial for guiding cancer treatment and has recently received great attention from researchers. However, the high complexity and heterogeneity of cancer gene regulatory networks limit the precition accuracy of existing deep learning models. To address this, we introduce a model called SCIS-CDG that utilizes Schur complement graph augmentation and independent subspace feature extraction techniques to effectively predict potential CDGs. Firstly, a random Schur complement strategy is adopted to generate two augmented views of gene network within a graph contrastive learning framework. Rapid randomization of the random Schur complement strategy enhances the model's generalization and its ability to handle complex networks effectively. Upholding the Schur complement principle in expectations promotes the preservation of the original gene network's vital structure in the augmented views. Subsequently, we employ feature extraction technology using multiple independent subspaces, each trained with independent weights to reduce inter-subspace dependence and improve the model's expressiveness. Concurrently, we introduced a feature expansion component based on the structure of the gene network to address issues arising from the limited dimensionality of node features. Moreover, it can alleviate the challenges posed by the heterogeneity of cancer gene networks to some extent. Finally, we integrate a learnable attention weight mechanism into the graph neural network (GNN) encoder, utilizing feature expansion technology to optimize the significance of various feature levels in the prediction task. Following extensive experimental validation, the SCIS-CDG model has exhibited high efficiency in identifying known CDGs and uncovering potential unknown CDGs in external datasets. Particularly when compared to previous conventional GNN models, its performance has seen significant improved. The code and data are publicly available at: https://github.com/mxqmxqmxq/SCIS-CDG.

Dynamic Reactive Spiking Graph Neural Network

Spiking Graph Neural Networks are emerging tools for analyzing graph data along with low energy consumption and certain biological fidelity. Existing methods directly integrate same-reactive spiking neurons into graph neural networks for processing propagated graphs. However, such same-reactive neurons are not biological-functionality enough compared to the brain's dynamic-reactive ones, limiting the model's expression. Meanwhile, insufficient long-range neighbor information can be excavated with the few-step propagated graph, restricting discrimination of graph spiking embeddings. Inspired by the dynamic cognition in the brain, we propose a Dynamic Reactive Spiking Graph Neural Network that can enhance model's expressive ability in higher biological fidelity. Specifically, we design dynamic reactive spiking neurons to process spiking graph inputs, which have unique optimizable thresholds to spontaneously explore dynamic reactive states between neurons. Moreover, discriminative graph positional spikes are learned and integrated adaptively into spiking outputs through our neurons, thereby exploring long-range neighbors more thoroughly. Finally, with the dynamic reactive mechanism and learnable positional integration, we can obtain a powerful and highly bio-fidelity model with low energy consumption. Experiments on various domain-related datasets can demonstrate the effectiveness of our model. Our code is available at https://github.com/hzhao98/DRSGNN.

Biases Mitigation and Expressiveness Preservation in Language Models: A Comprehensive Pipeline (Student Abstract)

Pre-trained language models (PLMs) have greatly transformed various downstream tasks, yet frequently display social biases from training data, raising fairness concerns. Recent efforts to debias PLMs come with limitations: they either fine-tune the entire parameters in PLMs, which is time-consuming and disregards the expressiveness of PLMs, or ignore the reintroducing biases from downstream tasks when applying debiased models to them. Hence, we propose a two-stage pipeline to mitigate biases from both internal and downstream contexts while preserving expressiveness in language models. Specifically, for the debiasing procedure, we resort to continuous prefix-tuning, not fully fine-tuning the PLM, in which we design a debiasing term for optimization and an alignment term to keep words’ relative distances and ensure the model's expressiveness. For downstream tasks, we perform causal intervention across different demographic groups for invariant predictions. Results on three GLUE tasks show our method alleviates biases from internal and downstream contexts, while keeping PLM expressiveness intact.

Prototype Bayesian Meta-Learning for Few-Shot Image Classification.

Meta-learning aims to leverage prior knowledge from related tasks to enable a base learner to quickly adapt to new tasks with limited labeled samples. However, traditional meta-learning methods have limitations as they provide an optimal initialization for all new tasks, disregarding the inherent uncertainty induced by few-shot tasks and impeding task-specific self-adaptation initialization. In response to this challenge, this article proposes a novel probabilistic meta-learning approach called prototype Bayesian meta-learning (PBML). PBML focuses on meta-learning variational posteriors within a Bayesian framework, guided by prototype-conditioned prior information. Specifically, to capture model uncertainty, PBML treats both meta-and task-specific parameters as random variables and integrates their posterior estimates into hierarchical Bayesian modeling through variational inference (VI). During model inference, PBML employs Laplacian estimation to approximate the integral term over the likelihood loss, deriving a rigorous upper-bound for generalization errors. To enhance the model's expressiveness and enable task-specific adaptive initialization, PBML proposes a data-driven approach to model the task-specific variational posteriors. This is achieved by designing a generative model structure that incorporates prototype-conditioned task-dependent priors into the random generation of task-specific variational posteriors. Additionally, by performing latent embedding optimization, PBML decouples the gradient-based meta-learning from the high-dimensional variational parameter space. Experimental results on benchmark datasets for few-shot image classification illustrate that PBML attains state-of-the-art or competitive performance when compared to other related works. Versatility studies demonstrate the adaptability and applicability of PBML in addressing diverse and challenging few-shot tasks. Furthermore, ablation studies validate the performance gains attributed to the inference and model components.

Explainability in subgraphs-enhanced Graph Neural Networks

Recently, subgraphs-enhanced Graph Neural Networks (SGNNs) have been introduced to enhance the expressive power of Graph Neural Networks (GNNs), which was proved to be not higher than the 1-dimensional Weisfeiler-Leman isomorphism test. The new paradigm suggests using subgraphs extracted from the input graph to improve the model's expressiveness, but the additional complexity exacerbates an already challenging problem in GNNs: explaining their predictions. In this work, we adapt PGExplainer, one of the most recent explainers for GNNs, to SGNNs. The proposed explainer accounts for the contribution of all the different subgraphs and can produce a meaningful explanation that humans can interpret. The experiments that we performed both on real and synthetic datasets show that our framework is successful in explaining the decision process of an SGNN on graph classification tasks.

Generative quantum learning of joint probability distribution functions

Modeling joint probability distributions is an important task in a wide variety of fields. One popular technique for this employs a family of multivariate distributions with uniform marginals called copulas. While the theory of modeling joint distributions via copulas is well understood, it gets practically challenging to accurately model real data with many variables. In this work, we design quantum machine learning algorithms to model copulas. We show that any copula can be naturally mapped to a multipartite maximally entangled state. A variational ansatz we christen as a `qopula' creates arbitrary correlations between variables while maintaining the copula structure starting from a set of Bell pairs for two variables, or GHZ states for multiple variables. As an application, we train a Quantum Generative Adversarial Network (QGAN) and a Quantum Circuit Born Machine (QCBM) using this variational ansatz to generate samples from joint distributions of two variables for historical data from the stock market. We demonstrate our generative learning algorithms on trapped ion quantum computers from IonQ for up to 8 qubits and show that our results outperform those obtained through equivalent classical generative learning. Further, we present theoretical arguments for exponential advantage in our model's expressivity over classical models based on communication and computational complexity arguments.

An Object Tracking Algorithm with Channel Reliability and Target Response Adaptation

In order to solve the problems of lower precision of target location in short-term occlusion and inaccurate of scale estimation of target in rotation by Spatial-Temporal Regularized Correlation Filters (STRCF), an object tracking algorithm with channel reliability and target response adaptation is proposed in this paper. In this algorithm, target response regularization is added to train target model. By updating the ideal target response function in the process of solving model, the target can be tracked again after being occluded for a short time. The reliability of each feature channel is evaluated by coefficient of channel reliability, which can improves the model's expression of the target. Scale filters can be trained in log-polar coordinates to improve the accuracy of scale estimation when target is rotating. The experimental results show that the proposed algorithm reduces 28.54 pixels in the average center position error and improves the average overlap rate by 22.8% compared with STRCF.

Potentiostatic Current Transient for Multiple Nucleation: A Limited-Diffusion Process Description

Potentiostatic current transient is widely used to analyze the diffusion controlled three-dimensional nucleation process. SH model has a satisfied performance for almost all potentiostatic testing. However, some deviations between experimental results and the SH model's expression at decreasing period were reported. In this paper, we describe the deviation as resulting from diffusion depth is limited in electrolytic cell. Through a stochastic model is proposed, we derive an equation to describe reaction near the electrode surface of solute ions and their migration out of the diffusion zone. The theoretical model describes a limited-diffusion process using hemispherical nucleus revealed in SH model and Brownian motion in MD model. This model give a detail description of multiple nucleation process during electrodeposition. Based on the descriptions, it can be seen that SH model is a limiting case of our model, as diffusion depth R →+∞. And the equations also reveal the relationship between SH model and our new model. In addition, our theoretical equations are powerful enough to reproduce the potentiostatic current transient curves obtained from gold electrodepositing process. This work advances earlier models a step further and gives an explanation for the deviations between experimental data and standard model's description.

A pattern-based approach for developing creativity applications supported by surface computing

An activity model for applications focused on creativity.Building blocks for constructing applications designed for creativity.By relying on our building blocks, applications were implemented in short time.Our building blocks prove the model's expressivity for describing creativity sessions. Whereas creativity tasks have traditionally been supported by conventional media and tools (such as paper, pens, scissors and glue); mobile phones, tablets and other devices based on interactive surfaces are increasingly been used as additional support. Large-sized multi-touch interactive surfaces appear as an interesting alternative for supporting creativity processes and for supporting synchronous collocated collaboration. However, they have mostly been usedfor visualization and navigation purposes. Their use as authoring means, which would be essential in creativity tasks, has only begun to be explored. Applications and platforms that have been developed in this area rely on low-level primitives for implementing representation of ideas and discussions. We have identified a significant gap between the level of development tools and the abstractions required by end-user applications that aim to support creativity processes using interactive surfaces. This gap makes it difficult for developers to build applications that provide richer, more flexible support for innovators working collaboratively around interactive surfaces. Based upon a thorough analysis of existing applications and user practices in the field, we have identified the key actions and interaction patterns that take place during collaborative creativity sessions. Thus, we propose ISCALI (Innovation Solutions Centered on Activities for Large-sized Interfaces), a model that can be used both for describing and for prescribing the role of multi-touch surfaces in collaborative creativity tasks. In accordance to Activity Centered Design, ISCALI comprises three major components: activities, actions and operations. The central activities within the processes of creativity comprise generation, organization and evaluation of ideas. Each of these activities encompasses sets of actions. Finally, several operation sets achieve the goal of each of the actions. Based upon our model, we designed a general architecture for collaborative creativity applications. This architecture addresses the development gap through a proposed set ofbuilding blocks. These building blocks implement the main interaction patterns needed for stimulating creativity tasks that rely on interactive surfaces. We have implemented prototypical versions of these building blocks, referred to as TOKAs (Touch Operations for Creative Activities), and have made them available to developers. Independent developers have implemented applications that facilitate the use of various methodologies that foster creativity and synchronous collocated collaboration. These developers have taken advantage of the availability of TOKAs. The implementation and use of TOKAs demonstrate ISCALI's expressivity for describing and guiding the development of applications that support collaborative creativity on top of interactive surfaces.

Extending social networks with delegation

Despite their enormous growth, current social networks lack a systematic approach to delegate rights – when an entity authorizes another to access the resources on its behalf. This paper proposes a delegation model based on socio-technical design and theory of cooperation and collaboration that best suits the requirements of social networks. The model is formulated through formal methods, designed using ontologies and implemented through Facebook APIs. The model's expressiveness is examined for overlapping policies of multiple users, its consistency is analyzed for conflicting and redundant policies and user acceptance testing is performed for acceptability. For social validity, the model is also compared with 27 previous delegation models with respect to socio-technical validity parameters derived from social principles already accepted in the human society.

Time Series Forecasting Based on Cloud Process Neural Network

AbstractTime series forecasting has been an important tool in many areas such as agriculture, finance, management, production or sales. In recent years, a large literature has evolved on the use of artificial neural networks (ANN) in time series forecasting. However conventional ANN is limited by its instantaneous synchronization inputs, it is difficult to express accumulative time effect and lacks certain processing ability for uncertainty factors (e.g., randomness, fuzziness ) hidden in time series. Thus a cloud process neural network (CPNN) model is put forward in the paper for time series forecasting. It combines cloud model's expression ability for uncertainty concepts and process neural network's dynamic signal processing method, converts quantitative time series inputs into multiple qualitative sub-cloud concepts, and then finds out the association rule between input and output variables through mining inherent law among multiple sub-clouds. For CPNN learning, this paper proposes a learning strateg...

A Method of Constructing the Fuel Efficiency Model Based on Quadratic Polynomial Regression

This paper deeply analyzes the data which is closely related to the fuel efficiency in the production database of the airline transportation. It verifies that quadratic polynomial regression model is the best fitting fuel efficiency model after many experiments. It constructs the quadratic polynomial regression model of the fuel efficiency and derives the model's expression using the least square method. And, an example shows the rationality of the model, and better results of testing. Above all, it obtains the influences’ degree of energy-saving and emission-reducing measures in civil aviation to the fuel efficiency, and provides a good theoretical support for further optimization of energy-saving and emission-reducing measures.

Electrooptical Study of an In-Plane Switching Mode Using a Uniaxial Medium Model

A new modeling method using a uniaxial medium model has been proposed to obtain a simple and quantitative expression for the transmittance of an in-plane switching (IPS) mode. In this model, the effective birefringence (Δneff) and the apparent optic axis rotation angle (φapp) are used instead of the original birefringence and the director profile, and Δneff changes when an electric field is applied to the IPS mode cell. The dependences of Δneff (measured as a retardation) and φapp on the applied voltage have been evaluated first by experiment. Δneff was reduced to 80% of the original value and φapp rotated 45° at the maximum transmittance (Tmax). It has been confirmed that the transmittance can be reproduced satisfactorily by employing the model's expression using the observed Δneff and φapp values. Theoretical simulation has also been executed and the results supported the experimental results. The simulation has also shown that both the Δneff–Δn ratio and φapp at Tmax are almost constant and do not depend on either the cell gap or liquid-crystal properties. Tmax can be easily calculated using the uniaxial medium model.