Advanced Learning Methods Research Articles

Dual-drive RUL prediction of gear transmission systems based on dynamic model and unsupervised domain adaption under zero sample

Gear transmission systems are key components in rotating machinery, and its remaining useful life (RUL) prediction can provide adequate leading time for well-timed maintenance. Existing RUL prediction models usually rely on sufficient training data. However, the acquisition of full life cycle dataset of gear transmission system is difficult or impossible. To solve the problem on the difficult startup and poor generalization of prediction model under zero environment, a dual-drive prediction method based on dynamic model of gear transmission system and unsupervised domain adaptation is proposed in this paper. Firstly, a dynamic model of gear transmission system and growth mechanism of local defect is established to generate full-life cycle simulation data. Then, the multi-scale modulation features are extracted based on simulated and measured data. Furthermore, multi-scale temporal convolution operations are introduced into dual-channel unsupervised domain adaptation model. Besides, a compound principle of reverse truncation and forward expansion principle is investigated to determine the first prediction time. Finally, the validity of the proposed model is verified by two kinds of gearbox data. Ablation experiments are carried out to evaluate the contribution of each module in proposed model. In addition, the effectiveness and generalization ability of proposed model are verified when compared with other advanced transfer learning methods.

Multimodal Learning for Mapping the Genotype-Phenotype Dynamics.

How complex phenotypes emerge from intricate gene expression patterns is a fundamental question in biology. Quantitative characterization of this relationship, however, is challenging due to the vast combinatorial possibilities and dynamic interplay between genotype and phenotype landscapes. Integrating high-content genotyping approaches such as single-cell RNA sequencing and advanced learning methods such as language models offers an opportunity for dissecting this complex relationship. Here, we present a computational integrated genetics framework designed to analyze and interpret the high-dimensional landscape of genotypes and their associated phenotypes simultaneously. We applied this approach to develop a multimodal foundation model to explore the genotype-phenotype relationship manifold for human transcriptomics at the cellular level. Analyzing this joint manifold showed a refined resolution of cellular heterogeneity, enhanced precision in phenotype annotating, and uncovered potential cross-tissue biomarkers that are undetectable through conventional gene expression analysis alone. Moreover, our results revealed that the gene networks are characterized by scale-free patterns and show context-dependent gene-gene interactions, both of which result in significant variations in the topology of the gene network, particularly evident during aging. Finally, utilizing contextualized embeddings, we investigated gene polyfunctionality which illustrates the multifaceted roles that genes play in different biological processes, and demonstrated that for VWF gene in endothelial cells. Overall, this study advances our understanding of the dynamic interplay between gene expression and phenotypic manifestation and demonstrates the potential of integrated genetics in uncovering new dimensions of cellular function and complexity.

Optimizing the model-prediction of date palm fronds-derived producer gas and third generation biodiesel powered dual-fuel engine by employing Bayesian-optimized Boosted Regression Trees for enhanced prognostics

The global issue concerning the dual problem of economic energy shortage as well as greenhouse gas emission needs urgent attention. Due to its abundance and carbon neutrality, waste biomass is a good energy conversion feedstock. This research suggests a unique method for using date palm frond waste biomass-derived producer gas and algal biodiesel as sustainable diesel engine fuels. The Bayesian-optimized neural networks and Bayesian-optimized Boosted Regression Trees were employed to optimize engine combustion and emission performance. This was further enhanced with the application of k-cross fold to prevent the model from overfitting during training. The working of dual-fuel engines is a complex phenomenon owing to the employment of two different types of fuel one is liquid (biodiesel + diesel) while the other one is gaseous (Producer gas). The present endeavor was undertaken to handle this with the application of advanced data learning methods. The engine was tested at different compression ratios, engine loads, and fuel injection pressures for brake thermal efficiency, diesel fuel substitution, and emission. Both the Bayesian-optimized neural networks and Bayesian-optimized Boosted Regression Trees could predict the engine output with good prediction efficiency with comparable performance, however, the Bayesian-optimized Boosted Regression Trees-based models were more robust with a coefficient of determination value in the range of 0.9617–0.9999 and mean squared errors as low as 0.001. The mean absolute percentage error was very low in the range of 0.0316–5.231. The Theil’s U2 values were in the range of 0.0029–0.0126.

Night Time Car Accident Prevention with AI Vision

Abstract: Night vision object detection with computer vision performs a vital function in diverse fields, when its dark, regular computer vision has trouble seeing things clearly. But because of new technology, we can now detect objects better in the dark. Our approach uses advanced learning methods and special sensors to help us see in darkness. By improving images, picking out important details, and using strong computer programs, we can spot objects accurately. After lots of testing, we've found our system works really well, with few mistakes and it's fast too. It's useful in surveillance, self-driving cars, security, and rescue missions, making dark places safer. So, basically, our system is like giving night vision to computers. It's like putting on special glasses that help you see in the dark. With this technology, we can keep an eye on things even when it's hard to see, whether it's for keeping people safe or making sure our cars can drive themselves without crashing.

Reducing Ophthalmic Health Disparities Through Transfer Learning: A Novel Application to Overcome Data Inequality.

Race disparities in the healthcare system and the resulting inequality in clinical data among different races hinder the ability to generate equitable prediction results. This study aims to reduce healthcare disparities arising from data imbalance by leveraging advanced transfer learning (TL) methods. We examined the ophthalmic healthcare disparities at a population level using electronic medical records data from a study cohort (N = 785) receiving care at an academic institute. Regression-based TL models were usesd, transferring valuable information from the dominant racial group (White) to improve visual field mean deviation (MD) rate of change prediction particularly for data-disadvantaged African American (AA) and Asian racial groups. Prediction results of TL models were compared with two conventional approaches. Disparities in socioeconomic status and baseline disease severity were observed among the AA and Asian racial groups. The TL approach achieved marked to comparable improvement in prediction accuracy compared to the two conventional approaches as evident by smaller mean absolute errors or mean square errors. TL identified distinct key features of visual field MD rate of change for each racial group. The study introduces a novel application of TL that improved reliability of the analysis in comparison with conventional methods, especially in small sample size groups. This can improve assessment of healthcare disparity and subsequent remedy approach. TL offers an equitable and efficient approach to mitigate healthcare disparities analysis by enhancing prediction performance for data-disadvantaged group.

Advanced learning methods in maritime education and training: A bibliometric analysis on the digitalization of education and modern trends

AbstractThe minimum requirements in maritime education and training (MET) are set by the International Convention on Standards of Training, Certification and Watchkeeping for Seafarers (STCW). First established in 1978, the STCW convention standardized educational procedures for seafarers. Since then, it has become the most influential regulatory document for maritime education. The STCW code had major revisions in 1995 and 2010. These amendments were in response to the recognized need to bring the Convention up to date. However, it has been over a decade since the latest amendments came into force, and over the past decade, new educational technologies and digital educational methods have developed significantly. Not only that, but technologies used onboard ships have also advanced considerably. Today, digitalization and the fourth industrial revolution are rapidly reshaping the maritime industry, as well as maritime education. These issues coupled with the disruption caused by the Covid pandemic increased the need for more advanced training tools and methods in MET. Concerns are rising regarding the need for more comprehensive rules that will better reflect today's educational needs and progress in the literature. This paper takes a first step toward addressing these issues by improving the understanding of the currently used advanced teaching methods in MET. This bibliometric review provides developments in the field from the perspective of educators and researchers. As a result, we discovered several narratives and literature gaps in MET. This article reviews the currently investigated subjects and topics, used educational tools and equipment, types of research, and main influencers in MET research. The field is still newly developing, but at a rapid pace. Newly developed tools and methods are increasing, but advances are also constrained because they are made to fit the decade‐old regulations.

Nonlinear learning method for local causal structures

Recent years have witnessed the proliferation of causal learning techniques, aimed at extracting the abundant causal relationships embedded within observational data. In many scenarios, our primary focus lies in predicting a single target variable. In such cases, it becomes both inefficient and unnecessary to learn an entire causal network through advanced global learning methods. To address this challenge, the concept of local causal learning has been introduced to identify the direct causes and effects of a target variable of interest. However, current algorithms exhibit limitations stemming from their reliance on conditional independence tests, which only consider the linear and pairwise relationships but ignore the ubiquitous nonlinear and multivariate causality, making them lose efficacy in practical scenarios. This paper takes significant strides toward facilitating the real-world applications for local causal learning. To identify the nonlinear relations, this paper discovers the Markov boundary (MB) through calculating the minimal conditional covariance operator in reproducing kernel Hilbert space. This approach establishes a theoretical equivalence between the solution and MB. Subsequently, a nonlinear scoring mechanism for structure learning is employed based on the selected subset, yielding the optimal local causal structure. A series of extensive experiments serves to underscore the superiority of the proposed method.

A lightweight dense relation network with attention for hyperspectral image few-shot classification

Deep learning methods have significantly progressed in hyperspectral image (HSI) classification. However, deep learning relies on large labeled data for training. The cost of labeling samples is enormous. In practical classification tasks, only a few labeled samples are available. A lightweight dense relation network with attention (LDA-RN) was presented for HSI few-shot classification. First, we design a 3D feature embedding module for spectral–spatial feature extraction. A light attention module (CSA), including the channel attention (CA) submodule and spatial attention (SA) submodule, is applied to reinforce essential features. Dense connections are used to fully utilize the feature information in each layer. Second, we propose a 2D relation learning module for classification by comparing the similarity between samples, using global average pooling to reduce the model parameters. Finally, meta-learning and fine-tuning training techniques are employed to enhance the model’s generalization ability and reduce the dependence on the labeled samples. Experimental results on three available HSI datasets suggest that the proposed LDA-RN outperforms existing advanced few-shot learning methods. For example, the experimental results show that the overall accuracy of the proposed method improves by 5.67% on the Univerisity of Pavia data set with only five labeled samples compared to the best method DCFSL.

Preface

The 2023 2nd International Conference on Machine Vision, Automatic Identification and Detection (MVAID 2023) took place in Hangzhou, China from April 14th to 16th, 2023, with the attendance of about 60 people.MVAID 2023 focused on research areas such as Computer Vision, Image Processing, Document Processing and Recognition, Advanced Learning Methods, Sensor Technique & Application, Image Forensics and Identification, etc. The main goal was to bring together basic scientists, researchers and developers engaged in machine vision, automatic identification and detection to bridge the gap between research needs and advanced technological solutions. MVAID 2023 provided an opportunity for researchers and related personnels to present their recent theoretical developments and practical applications as well as to discuss unsolved problems. It also fostered research and development in machine vision, automatic identification and detection.The Program of the Conference was divided into three main topics. First day of the Conference was devoted to registration for all the participants. The second day was focused on keynote speech, oral and poster presentation, as well as discussion on the talks and the whole meeting. Academic investigation took place during the third day of the Conference.We would like to highlight the main developments presented at the Conference and introduce the prominent speakers. The keynote speaker of the Conference Professor Yulin Wang (Wuhan University, China) presented a report on Drone: From Bionic Flight to Brain Like Autonomous Navigation. He mainly works for image and video processing, digital rights management, information security, intelligent system, e-commerce, IoT, code clone and so on. Professor Cairong Zhao (Tongji University, China) demonstrated his speech on Occluded Person Re-Identification, in which he elaborated on the four key issues caused by occlusion, namely positional displacement, scale displacement, noise information, and missing information, and discuss their solutions, and analyzed and discussed the future development trends of occluded person re-identification.Finally, we would like to thank the members of the Technical Program Committee for giving an opportunity to share scientific ideas and new developments at the Conference as well as Hangzhou Dianzi University and Politechnika Koszalinska for their friendly support. Last but not the least, our special acknowledgement also goes to the editors and staff of Journal of Physics: Conference Series, for their kindness and assistance in publishing this volume.The Committee of MVAID 2023List of Committee Member is available in this Pdf.

Contrastive Learning with the Feature Reconstruction Amplifier

Contrastive learning has emerged as one of the most promising self-supervised methods. It can efficiently learn the transferable representations of samples through the instance-level discrimination task. In general, the performance of the contrastive learning method can be further improved by projecting the transferable high-dimensional representations into the low-dimensional feature space. This is because the model can learn more abstract discriminative information. However, when low-dimensional features cannot provide sufficient discriminative information to the model (e.g., the samples are very similar to each other), the existing contrastive learning method will be limited to a great extent. Therefore, in this paper, we propose a general module called the Feature Reconstruction Amplifier (FRA) for adding additional high-dimensional feature information to the model. Specifically, FRA reconstructs the low-dimensional feature embeddings with Gaussian noise vectors and projects them to a high-dimensional reconstruction space. In this reconstruction space, we can add additional feature information through the designed loss. We have verified the effectiveness of the module itself through exhaustive ablation experiments. In addition, we perform linear evaluation and transfer learning on five common visual datasets, the experimental results demonstrate that our method is superior to recent advanced contrastive learning methods.

Digital twin aided adversarial transfer learning method for domain adaptation fault diagnosis

Machine health management has become the focus of equipment monitoring upgrading with the advance of digital twin (DT). The DT model is able to generate system performance data that is close to reality, which opens a new way for the cyber-physical integration of equipment monitoring. Furthermore, it also provides a significant opportunity for mechanical fault diagnosis when the collected fault signals are insufficient. In this paper, a DT aided intelligent fault diagnosis model is proposed for triplex pump. Specifically, the simulation model of the triplex pump is built by Simscape in MATLAB, and the measured simulation data is continuously updated to construct the DT model. Then a novel transfer learning model based on domain-adversarial strategy and Wasserstein distance is present and trained by the source domain data which generated from the DT model. Next, the opening pressure of the triplex pump is controlled to simulate different working conditions, so as to achieve feature transfer and fault diagnosis for the DT model. The experimental results show that the proposed method is effective and superior to other advanced transfer learning methods.

Heterogeneous graph neural network with semantic-aware differential privacy guarantees

Most social networks can be modeled as heterogeneous graphs. Recently, advanced graph learning methods exploit the rich node properties and topological relationships for downstream tasks. That means that more private information is embedded in the representation. However, the existing privacy-preserving methods only focus on protecting the single type of node attributes or relationships, which neglect the significance of high-order semantic information. To address this issue, we propose a novel Heterogeneous graph neural network with Semantic-aware Differential privacy Guarantees named HeteSDG, which provides a double privacy guarantee and performance trade-off in terms of both graph features and topology. In particular, we first reveal the privacy leakage in heterogeneous graphs and define a membership inference attack with a semantic enhancement (MIS) that will improve the means of member inference attacks by obtaining side background knowledge through semantics. Then we design a two-stage mechanism, which includes the feature attention personalized mechanism and the topology gradient perturbation mechanism, where the privacy-preserving technologies are based on differential privacy. These mechanisms will defend against MIS and provide stronger interpretation, but simultaneously bring in noise for representation learning. To better balance the noise perturbation and learning performance, we utilize a bi-level optimization pattern to allocate a suitable privacy budget for the above two modules. Our experiments on four public benchmarks conduct performance experiments, ablation studies, inference attack verification, etc. The results show the privacy protection capability and generalization of HeteSDG.

Clinical applications of deep learning in breast MRI.

Deep learning (DL) is one of the most powerful data-driven machine-learning techniques in artificial intelligence (AI). It can automatically learn from raw data without manual feature selection. DL models have led to remarkable advances in data extraction and analysis for medical imaging. Magnetic resonance imaging (MRI) has proven useful in delineating the characteristics and extent of breast lesions and tumors. This review summarizes the current state-of-the-art applications of DL models in breast MRI. Many recent DL models were examined in this field, along with several advanced learning approaches and methods for data normalization and breast and lesion segmentation. For clinical applications, DL-based breast MRI models were proven useful in five aspects: diagnosis of breast cancer, classification of molecular types, classification of histopathological types, prediction of neoadjuvant chemotherapy response, and prediction of lymph node metastasis. For subsequent studies, further improvement in data acquisition and preprocessing is necessary, additional DL techniques in breast MRI should be investigated, and wider clinical applications need to be explored.

Low-Rank Constrained Attention-Enhanced Multiple Spatial–Spectral Feature Fusion for Small Sample Hyperspectral Image Classification

Hyperspectral images contain rich features in both spectral and spatial domains, which bring opportunities for accurate recognition of similar materials and promote various fine-grained remote sensing applications. Although deep learning models have been extensively investigated in the field of hyperspectral image classification (HSIC) tasks, classification performance is still limited under small sample conditions, and this has been a longstanding problem. The features extracted by complex network structures with large model size are redundant to some extent and prone to overfitting. This paper proposes a low-rank constrained attention-enhanced multiple feature fusion network (LAMFN). Firstly, factor analysis is used to extract very few components that can describe the original data using covariance information to perform spectral feature preprocessing. Then, a lightweight attention-enhanced 3D convolution module is used for deep feature extraction, and the position-sensitive information is supplemented using a 2D coordinate attention. The above widely varying spatial–spectral feature groups are fused through a simple composite residual structure. Finally, low-rank second-order pooling is adopted to enhance the convolutional feature selectivity and achieve classification. Extensive experiments were conducted on four representative hyperspectral datasets with different spatial–spectral characteristics, namely Indian Pines (IP), Pavia Center (PC), Houston (HU), and WHU-HongHu (WHU). The contrast methods include several advanced models proposed recently, including residual CNNs, attention-based CNNs, and transformer-based models. Using only five samples per class for training, LAMFN achieved overall accuracies of 78.15%, 97.18%, 81.35%, and 87.93% on the above datasets, which has an improvement of 0.82%, 1.12%, 1.67%, and 0.89% compared to the second-best model. The running time of LAMFN is moderate. For example, the training time of LAMFN on the WHU dataset was 29.1 s, and the contrast models ranged from 3.0 s to 341.4 s. In addition, ablation experiments and comparisons with some advanced semi-supervised learning methods further validated the effectiveness of the proposed model designs.

Remaining Useful Life Prediction of Bearings Based on Convolution Attention Mechanism and Temporal Convolution Network

The prediction of the remaining useful life (RUL) of bearings is of great significance for reducing cost and increasing efficiency of mechanical equipment and ensuring healthy operation. Most of the traditional RUL prediction methods based on deep learning only analyze single sensor data, which makes the prediction accuracy fluctuate greatly and the reliability is low, and the contributions of data collected by different sensors to RUL prediction are inconsistent, which makes the low utilization of data lead to poor prediction results. To solve the above problems, a new RUL prediction framework with convolutional attention mechanism and temporal convolutional network (CAMTCN) is proposed. Firstly, the vibration signals collected by multi-sensor are normalized and processed directly as the input of the network. Secondly, an Efficient Adaptive Shrinkage (EAS) model is designed to adaptively eliminate noise interference and improve the accuracy of RUL prediction; Finally, we use the attention mechanism to design the convolutional attention subnet and combine it with the TCN to obtain the contribution degree of bearing degradation features collected by different sensors. Adaptive weight allocation captures more important degradation features to achieve end-to-end RUL prediction. Experiments were conducted on IEEE PHM 2012 dataset and XJTU-SY dataset respectively, and compared with some advanced in-depth learning methods. The experimental results show that this method can predict RUL more accurately on different bearing data sets, and has good generalization.

RESEARCH ON THE PRINCIPLES OF ARTIFICIAL NEURAL NETWORK CONSTRUCTION AND PARAMETER MODELING

Studies show that neural networks with multilayer direct propagation have be-come more widespread, which is due to the simplicity of their implementation, the availability of advanced learning methods, and the parallel execution of calculations. In these networks, the computational complexity increases proportionally to the square of the data size, which leads to a decrease in performance and an increase in hardware costs. The main problems of synthesis of artificial neural networks, criteria for evaluating the effectiveness of artificial neural networks, issues of choosing layers and the number of neurons in each layer are investigated. In the work, the mean square error, mean absolute error and mean absolute percentage error criteria are considered for evaluating the efficiency of neural networks. In general, one hidden layer of a neu-ral network is sufficient for most problems, two hidden layers can represent a function of any shape, and adding a third and more layers results in very little improvement in network performance. Expressions are given for the total number of parameters de-pending on the number, and the number of inputs from the binary sequence, as well as the minimum and maximum values of the training examples depending on the total number of parameters and the number of outputs. The studies have shown that the se-lection of the statistical model efficiency evaluation criteria depends on the problem to be solved. The results of the simulations performed using the MATLAB package are presented.

Modeling and feature assessment of the sleep quality among chronic kidney disease patients

Chronic Kidney Disease (CKD) is a progressive and irreversible loss of kidney function. Data mining concepts may be used in assessing and predicting CKD-related issues to obtain hidden clinical information for a reliable and effective decision-making process. These advanced learning methods would identify the relationships and patterns that will help classify factors that affect the poor sleep quality of CKD patients. Poor sleep quality is a critical issue for CKD individuals, negatively affecting immunity, cognitive functions, and emotional demonstrations. This study aims to find the factors affecting the sleep quality of CKD patients. Decision tree-based methods are used to identify the impact of each feature to predict sleep quality. The predictive results are compared with different classification models as well. Furthermore, two re-sampling techniques, Synthetic Minority Oversampling and Random Oversampling, are also used to reduce the impact of the imbalanced nature of the data set. We further discuss how these results agree with the clinically relevant features determined by the physicians.

Advancements in deep structured learning based medical image interpretation

Deep learning is a fast flattering highest development in the machine learning technique leading to augment performance in diverse medical applications. The tributary artificial intelligence is rapidly becoming metamorphic in the domain of Healthcare Technology rendering facility to analyze the medical data with high rapidity and exactness. Due to the overwhelming development of artificial intelligence (AI), the contemporary encroachments of deep learning employing advanced and futuristic deep learning methods has become a vivacious research domain for processing medical data in the health care industries. This paper presents a complete analysis about the deep structured learning based methods used in medical image analysis and also elucidate about the advancement of convolutional neural network which is a major class of deep neural networks used in medical image processing. It also gives insight about the fundamentals of deep learning methods and reviews in image interpretation. It also explains about the implementation of an integrated framework for medical image synthesis combining Speckle GAN and Di-Conv-AE-Net.

Inverse design of spontaneous parametric downconversion for generation of high-dimensional qudits

Spontaneous parametric downconversion (SPDC) in quantum optics is an invaluable resource for the realization of high-dimensional qudits with spatial modes of light. One of the main open challenges is how to directly generate a desirable qudit state in the SPDC process. This problem can be addressed through advanced computational learning methods; however, due to difficulties in modeling the SPDC process by a fully differentiable algorithm, progress has been limited. Here, we overcome these limitations and introduce a physically constrained and differentiable model, validated against experimental results for shaped pump beams and structured crystals, capable of learning the relevant interaction parameters in the process. We avoid any restrictions induced by the stochastic nature of our physical model and integrate the dynamic equations governing the evolution under the SPDC Hamiltonian. We solve the inverse problem of designing a nonlinear quantum optical system that achieves the desired quantum state of downconverted photon pairs. The desired states are defined using either the second-order correlations between different spatial modes or by specifying the required density matrix. By learning nonlinear photonic crystal structures as well as different pump shapes, we successfully show how to generate maximally entangled states. Furthermore, we simulate all-optical coherent control over the generated quantum state by actively changing the profile of the pump beam. Our work can be useful for applications such as novel designs of high-dimensional quantum key distribution and quantum information processing protocols. In addition, our method can be readily applied for controlling other degrees of freedom of light in the SPDC process, such as spectral and temporal properties, and may even be used in condensed-matter systems having a similar interaction Hamiltonian.

Multi-Scale Part-Based Syndrome Classification of 3D Facial Images

Identification and delineation of craniofacial characteristics support the clinical and molecular diagnosis of genetic syndromes. Deep learning (DL) frameworks for syndrome identification from 2D facial images are trained on large clinical datasets using standard convolutional neural networks for classification. In contrast, despite the increased availability of 3D scanners in clinical setups, similar frameworks remain absent for 3D facial photographs. The main challenges involve working with smaller datasets and the need for DL operations applicable to 3D geometric data. Therefore, to date, most 3D methods refrain from working across multiple syndromic groups and/or are solely based on traditional machine learning. The first contribution of this work is the use of geometric deep learning with spiral convolutions in a triplet-loss architecture. This geometric encoding (GE) learns a lower dimensional metric space from 3D facial data that is used as input to linear discriminant analysis (LDA) performing multiclass classification. Benchmarking is done against principal component analysis (PCA), a common technique in 3D facial shape analysis, and related work based on 65 distinct 3D facial landmarks as input to LDA. The second contribution of this work involves a part-based implementation to 3D facial shape analysis and multi-class syndrome classification, and this is applied to both GE and PCA. Based on 1,786 3D facial photographs of controls and individuals from 13 different syndrome classes, a five-fold cross-validation was used to investigate both contributions. Results indicate that GE performs better than PCA as input to LDA, and this especially so for more compact (lower dimensional) spaces. In addition, a part-based approach increases performance significantly for both GE and PCA, with a more significant improvement for the latter. I.e., this contribution enhances the power of the dataset. Finally, and interestingly, according to ablation studies within the part-based approach, the upper lip is the most distinguishing facial segment for classifying genetic syndromes in our dataset, which follows clinical expectation. This work stimulates an enhanced use of advanced part-based geometric deep learning methods for 3D facial imaging in clinical genetics.