Amount Of Training Data Research Articles

VolTeMorph: Real‐time, Controllable and Generalizable Animation of Volumetric Representations

AbstractThe recent increase in popularity of volumetric representations for scene reconstruction and novel view synthesis has put renewed focus on animating volumetric content at high visual quality and in real‐time. While implicit deformation methods based on learned functions can produce impressive results, they are ‘black boxes’ to artists and content creators, they require large amounts of training data to generalize meaningfully, and they do not produce realistic extrapolations outside of this data. In this work, we solve these issues by introducing a volume deformation method which is real‐time even for complex deformations, easy to edit with off‐the‐shelf software and can extrapolate convincingly. To demonstrate the versatility of our method, we apply it in two scenarios: physics‐based object deformation and telepresence where avatars are controlled using blendshapes. We also perform thorough experiments showing that our method compares favourably to both volumetric approaches combined with implicit deformation and methods based on mesh deformation.

Theory guided Lagrange programming neural network for subsurface flow problems

A deep learning model can perform efficient uncertainty quantification (UQ) for reservoir flow with uncertain model parameters, but usually requires large amounts of training data to ensure accuracy. However, the cost of obtaining large amounts of data is prohibitive, and the performance will deteriorate if sufficient training data is lacking. Alternatively, more interpretable neural networks with embedded physical laws have recently been used to solve partial differential equations as well as to solve UQ problems. This approach has received a lot of attention due to its low data volume requirements and its adherence to the laws of physics during the training process. In this paper, we propose a theory-guided framework based on a bilevel programming model with hard constraints to embed physical meaning in the model. Theory guided Lagrange programming neural network (TGLPNN) combines the method of Lagrange programming neural network approach where physical laws such as stochastic partial differential equations and boundary conditions are incorporated into the training process of a convolutional neural network. At the same time, the upper-level variables are iteratively optimized. The method based on Lagrange programming neural network inherently embeds physical laws in the network. Practical applications have shown that TGLPNN can provide higher prediction accuracy compared to state-of-the-art physics-driven methods and improved efficiency compared to numerical methods.

A comparative analysis of different augmentations for brain images.

Deep learning (DL) requires a large amount of training data to improve performance and prevent overfitting. To overcome these difficulties, we need to increase the size of the training dataset. This can be done by augmentation on a small dataset. The augmentation approaches must enhance the model's performance during the learning period. There are several types of transformations that can be applied to medical images. These transformations can be applied to the entire dataset or to a subset of the data, depending on the desired outcome. In this study, we categorize data augmentation methods into four groups: Absent augmentation, where no modifications are made; basic augmentation, which includes brightness and contrast adjustments; intermediate augmentation, encompassing a wider array of transformations like rotation, flipping, and shifting in addition to brightness and contrast adjustments; and advanced augmentation, where all transformation layers are employed. We plan to conduct a comprehensive analysis to determine which group performs best when applied to brain CT images. This evaluation aims to identify the augmentation group that produces the most favorable results in terms of improving model accuracy, minimizing diagnostic errors, and ensuring the robustness of the model in the context of brain CT image analysis.

MousiPLIER: A Mouse Pathway-Level Information Extractor Model.

High-throughput gene expression profiling measures individual gene expression across conditions. However, genes are regulated in complex networks, not as individual entities, limiting the interpretability of gene expression data. Machine learning models that incorporate prior biological knowledge are a powerful tool to extract meaningful biology from gene expression data. Pathway-level information extractor (PLIER) is an unsupervised machine learning method that defines biological pathways by leveraging the vast amount of published transcriptomic data. PLIER converts gene expression data into known pathway gene sets, termed latent variables (LVs), to substantially reduce data dimensionality and improve interpretability. In the current study, we trained the first mouse PLIER model on 190,111 mouse brain RNA-sequencing samples, the greatest amount of training data ever used by PLIER. We then validated the mousiPLIER approach in a study of microglia and astrocyte gene expression across mouse brain aging. mousiPLIER identified biological pathways that are significantly associated with aging, including one latent variable (LV41) corresponding to striatal signal. To gain further insight into the genes contained in LV41, we performed k-means clustering on the training data to identify studies that respond strongly to LV41. We found that the variable was relevant to striatum and aging across the scientific literature. Finally, we built a Web server (http://mousiplier.greenelab.com/) for users to easily explore the learned latent variables. Taken together, this study defines mousiPLIER as a method to uncover meaningful biological processes in mouse brain transcriptomic studies.

Training a Dataset Simulated Using RGB Images for an End-to-End Event-Based DoLP Recovery Network

Event cameras are bio-inspired neuromorphic sensors that have emerged in recent years, with advantages such as high temporal resolutions, high dynamic ranges, low latency, and low power consumption. Event cameras can be used to build event-based imaging polarimeters, overcoming the limited frame rates and low dynamic ranges of existing systems. Since events cannot provide absolute brightness intensity in different angles of polarization (AoPs), degree of linear polarization (DoLP) recovery in non-division-of-time (non-DoT) event-based imaging polarimeters is an ill-posed problem. Thus, we need a data-driven deep learning approach. Deep learning requires large amounts of data for training, and constructing a dataset for event-based non-DoT imaging polarimeters requires significant resources, scenarios, and time. We propose a method for generating datasets using simulated polarization distributions from existing red–green–blue images. Combined with event simulator V2E, the proposed method can easily construct large datasets for network training. We also propose an end-to-end event-based DoLP recovery network to solve the problem of DoLP recovery using event-based non-DoT imaging polarimeters. Finally, we construct a division-of-time event-based imaging polarimeter simulating an event-based four-channel non-DoT imaging polarimeter. Using real-world polarization events and DoLP ground truths, we demonstrate the effectiveness of the proposed simulation method and network.

Global sensitivity analysis with limited data via sparsity-promoting D-MORPH regression: Application to char combustion

In uncertainty quantification, variance-based global sensitivity analysis quantitatively determines the effect of each input random variable on the output by partitioning the total output variance into contributions from each input. However, computing conditional expectations can be prohibitively costly when working with expensive-to-evaluate models. Surrogate models can accelerate this, yet their accuracy depends on the quality and quantity of training data, which is expensive to generate (experimentally or computationally) for complex engineering systems. Thus, methods that work with limited data are desirable. We propose a diffeomorphic modulation under observable response preserving homotopy (D-MORPH) regression to train a polynomial dimensional decomposition surrogate of the output that minimizes the number of training data. The new method first computes a sparse Lasso solution and uses it to define the cost function. A subsequent D-MORPH regression minimizes the difference between the D-MORPH and Lasso solution. The resulting D-MORPH based surrogate is more robust to input variations and more accurate with limited training data. We illustrate the accuracy and computational efficiency of the new surrogate for global sensitivity analysis using mathematical functions and an expensive-to-simulate model of char combustion. The new method is highly efficient, requiring only 15% of the training data compared to conventional regression.

Versatile multiple object tracking in sparse 2D/3D videos via deformable image registration.

Tracking body parts in behaving animals, extracting fluorescence signals from cells embedded in deforming tissue, and analyzing cell migration patterns during development all require tracking objects with partially correlated motion. As dataset sizes increase, manual tracking of objects becomes prohibitively inefficient and slow, necessitating automated and semi-automated computational tools. Unfortunately, existing methods for multiple object tracking (MOT) are either developed for specific datasets and hence do not generalize well to other datasets, or require large amounts of training data that are not readily available. This is further exacerbated when tracking fluorescent sources in moving and deforming tissues, where the lack of unique features and sparsely populated images create a challenging environment, especially for modern deep learning techniques. By leveraging technology recently developed for spatial transformer networks, we propose ZephIR, an image registration framework for semi-supervised MOT in 2D and 3D videos. ZephIR can generalize to a wide range of biological systems by incorporating adjustable parameters that encode spatial (sparsity, texture, rigidity) and temporal priors of a given data class. We demonstrate the accuracy and versatility of our approach in a variety of applications, including tracking the body parts of a behaving mouse and neurons in the brain of a freely moving C. elegans. We provide an open-source package along with a web-based graphical user interface that allows users to provide small numbers of annotations to interactively improve tracking results.

Machine Learning-Based Hand Pose Generation Using a Haptic Controller

In this study, we present a novel approach to derive hand poses from data input via a haptic controller, leveraging machine learning techniques. The input values received from the haptic controller correspond to the movement of five fingers, each assigned a value between 0.0 and 1.0 based on the applied pressure. The wide array of possible finger movements requires a substantial amount of motion capture data, making manual data integration difficult. This challenge is primary due to the need to process and incorporate large volumes of diverse movement information. To tackle this challenge, our proposed method automates the process by utilizing machine learning algorithms to convert haptic controller inputs into hand poses. This involves training a machine learning model using supervised learning, where hand poses are matched with their corresponding input values, and subsequently utilizing this trained model to generate hand poses in response to user input. In our experiments, we assessed the accuracy of the generated hand poses by analyzing the angles and positions of finger joints. As the quantity of training data increased, the margin of error decreased, resulting in generated poses that closely emulated real-world hand movements.

Towards Increasing Performance Rate with Minimized Dataset in ECG-based Human Identification

This paper presents an ECG-based human identification that aims to increase the prediction accuracy with a minimized training dataset using a Block Dense CNN (BDCNN) architecture. This study specifically examines the effect of increasing the number of individuals, while decreasing the amount of training data on the performance of the network. Furthermore, the performance of the proposed network with multisession ECG recordings is also investigated. Comprehensive experimentation on nine ECG public databases that include normal and various abnormal ECG beats has been carried out. Performance of the network is evaluated using metrics such as prediction accuracy, false rejection ratio (FRR), false acceptance ratio (FAR), computational complexity, and training time, and the results are then compared to the results of a residual network. Experimental results show that, not only in a single session but also in multisession ECG recordings, the proposed network proficiently overbears the load of increased persons with decreased training dataset with about a 5.06% increase in identification rate as compared to its close counterpart. Therefore, the proposed BDCNN can be a potential framework for ECG-based human identification systems.

Cross-domain structural damage identification using transfer learning strategy

In damage identification of civil structures, training deep neural networks (DNNs) often requires a large volume of annotated training data. However, artificial damage to real-world structures is always forbidden, resulting in insufficient data for training. Transfer learning provides a solution that allows structural damage information in a data-rich domain to be transferred and shared as the prior knowledge to a data-scarce domain, thereby indirectly augmenting available training data in the latter domain. To this end, a multi-task transfer learning strategy has been adopted for the purpose of achieving cross-domain damage identification between a plane frame in the source domain and a three-dimensional frame in the target domain. The strategy can share the learning knowledge from the domain with sufficient training data to the target domain with insufficient data. Thereby, the DNN (named DNN#2) in the target domain can be trained on a small amount of training data. Owing to their ability in data feature extraction, stacked auto-encoders (SAEs) are used to construct the desirable DNN#1 and DNN#2 corresponding to different damage identification tasks, named Task#1 and Task#2, in the two domains. The two DNNs share the hidden layers in order to share damage feature information between the source and target domains. The training datasets of the two domains are first used to jointly pre-train the auto-encoders’ parameters in an unsupervised learning way. Afterwards, a supervised fine-tuning step is carried out to retraining the entire SAEs for better performance. By these means, Task#2 receives some prior knowledge from Task#1, and thus, it is accomplished on limited training data when Task#1 is synchronously achieved. The analysis results demonstrate that the proposed multi-task learning strategy requires only a single training session to simultaneously realize the cross-domain damage identification of two different steel frames.

Leveraging SO(3)-steerable convolutions for pose-robust semantic segmentation in 3D medical data

Convolutional neural networks (CNNs) allow for parameter sharing and translational equivariance by using convolutional kernels in their linear layers. By restricting these kernels to be SO(3)-steerable, CNNs can further improve parameter sharing and equivariance. These equivariant convolutional layers have several advantages over standard convolutional layers, including increased robustness to unseen poses, smaller network size, and improved sample efficiency. Despite this, most segmentation networks used in medical image analysis continue to rely on standard convolutional kernels. In this paper, we present a new family of segmentation networks that use equivariant voxel convolutions based on spherical harmonics. These SE(3)-equivariant volumetric segmentation networks, which are robust to data poses not seen during training, do not require rotation-based data augmentation during training. In addition, we demonstrate improved segmentation performance in MRI brain tumor and healthy brain structure segmentation tasks, with enhanced robustness to reduced amounts of training data and improved parameter efficiency. Code to reproduce our results, and to implement the equivariant segmentation networks for other tasks is available at <a href='http://github.com/SCAN-NRAD/e3nn_Unet'>http://github.com/SCAN-NRAD/e3nn_Unet</a>.

Real-Time Microgrid Energy Scheduling Using Meta-Reinforcement Learning

With the rapid development of renewable energy and the increasing maturity of energy storage technology, microgrids are quickly becoming popular worldwide. The stochastic scheduling problem of microgrids can increase operational costs and resource wastage. In order to reduce operational costs and optimize resource utilization efficiency, the real-time scheduling of microgrids becomes particularly important. After collecting extensive data, reinforcement learning (RL) can provide good strategies. However, it cannot make quick and rational decisions in different environments. As a method with generalization ability, meta-learning can compensate for this deficiency. Therefore, this paper introduces a microgrid scheduling strategy based on RL and meta-learning. This method can quickly adapt to different environments with a small amount of training data, enabling rapid energy scheduling policy generation in the early stages of microgrid operation. This paper first establishes a microgrid model, including components such as energy storage, load, and distributed generation (DG). Then, we use a meta-reinforcement learning framework to train the initial scheduling strategy, considering the various operational constraints of the microgrid. The experimental results show that the MAML-based RL strategy has advantages in improving energy utilization and reducing operational costs in the early stages of microgrid operation. This research provides a new intelligent solution for microgrids’ efficient, stable, and economical operation in their initial stages.

Large-scale 3D non-Cartesian coronary MRI reconstruction using distributed memory-efficient physics-guided deep learning with limited training data.

To enable high-quality physics-guided deep learning (PG-DL) reconstruction of large-scale 3D non-Cartesian coronary MRI by overcoming challenges of hardware limitations and limited training data availability. While PG-DL has emerged as a powerful image reconstruction method, its application to large-scale 3D non-Cartesian MRI is hindered by hardware limitations and limited availability of training data. We combine several recent advances in deep learning and MRI reconstruction to tackle the former challenge, and we further propose a 2.5D reconstruction using 2D convolutional neural networks, which treat 3D volumes as batches of 2D images to train the network with a limited amount of training data. Both 3D and 2.5D variants of the PG-DL networks were compared to conventional methods for high-resolution 3D kooshball coronary MRI. Proposed PG-DL reconstructions of 3D non-Cartesian coronary MRI with 3D and 2.5D processing outperformed all conventional methods both quantitatively and qualitatively in terms of image assessment by an experienced cardiologist. The 2.5D variant further improved vessel sharpness compared to 3D processing, and scored higher in terms of qualitative image quality. PG-DL reconstruction of large-scale 3D non-Cartesian MRI without compromising image size or network complexity is achieved, and the proposed 2.5D processing enables high-quality reconstruction with limited training data.

Image Fusion Method Based on Snake Visual Imaging Mechanism and PCNN.

The process of image fusion is the process of enriching an image and improving the image's quality, so as to facilitate the subsequent image processing and analysis. With the increasing importance of image fusion technology, the fusion of infrared and visible images has received extensive attention. In today's deep learning environment, deep learning is widely used in the field of image fusion. However, in some applications, it is not possible to obtain a large amount of training data. Because some special organs of snakes can receive and process infrared information and visible information, the fusion method of infrared and visible light to simulate the visual mechanism of snakes came into being. Therefore, this paper takes into account the perspective of visual bionics to achieve image fusion; such methods do not need to obtain a significant amount of training data. However, most of the fusion methods for simulating snakes face the problem of unclear details, so this paper combines this method with a pulse coupled neural network (PCNN). By studying two receptive field models of retinal nerve cells, six dual-mode cell imaging mechanisms of rattlesnakes and their mathematical models and the PCNN model, an improved fusion method of infrared and visible images was proposed. For the proposed fusion method, eleven groups of source images were used, and three non-reference image quality evaluation indexes were compared with seven other fusion methods. The experimental results show that the improved algorithm proposed in this paper is better overall than the comparison method for the three evaluation indexes.

Corrosion segmentation method of concrete drainage pipes based on point transformer

Deep learning based Closed Circuit Television (CCTV) image detection has achieved remarkable accuracy in identifying and segmenting concrete pipeline damages. However, the lack of in-depth information and requirement for large amounts of training data are two major drawbacks. Therefore, this study proposes a point cloud data segmentation method that incorporates depth information to address the issue of insufficient training data for depth models, enabling efficient and accurate segmentation. Firstly, 3D simulation technology is employed to construct a simulation dataset. Secondly, an optimized point transformation model is established to identify and segment the point cloud data. The proposed method achieved a test accuracy of 94.36 % by optimizing the network structure and the training strategy. The Mean Intersection over Union (MIoU) increased by 3.62 % and 10.3 %, respectively, compared with the classic PointNet++ and before adding simulation data, reaching 86.31 %. These results demonstrate that the proposed method exhibits high precision defect detection capability on three-dimensional point clouds of drainage pipelines.

Sample Size Effect on Musculoskeletal Segmentation: How Low Can We Go?

Convolutional Neural Networks have emerged as a predominant tool in musculoskeletal medical image segmentation. It enables precise delineation of bone and cartilage in medical images. Recent developments in image processing and network architecture desire a reevaluation of the relationship between segmentation accuracy and the amount of training data. This study investigates the minimum sample size required to achieve clinically relevant accuracy in bone and cartilage segmentation using the nnU-Net methodology. In addition, the potential benefit of integrating available medical knowledge for data augmentation, a largely unexplored opportunity for data preprocessing, is investigated. The impact of sample size on the segmentation accuracy of the nnU-Net is studied using three distinct musculoskeletal datasets, including both MRI and CT, to segment bone and cartilage. Further, the use of model-informed augmentation is explored on two of the above datasets by generating new training samples implementing a shape model-informed approach. Results indicate that the nnU-Net can achieve remarkable segmentation accuracy with as few as 10–15 training samples on bones and 25–30 training samples on cartilage. Model-informed augmentation did not yield relevant improvements in segmentation results. The sample size findings challenge the common notion that large datasets are necessary to obtain clinically relevant segmentation outcomes in musculoskeletal applications.

Labelling with dynamics: A data-efficient learning paradigm for medical image segmentation

The success of deep learning on image classification and recognition tasks has led to new applications in diverse contexts, including the field of medical imaging. However, two properties of deep neural networks (DNNs) may limit their future use in medical applications. The first is that DNNs require a large amount of labeled training data, and the second is that the deep learning-based models lack interpretability. In this paper, we propose and investigate a data-efficient framework for the task of general medical image segmentation. We address the two aforementioned challenges by introducing domain knowledge in the form of a strong prior into a deep learning framework. This prior is expressed by a customized dynamical system. We performed experiments on two different datasets, namely JSRT and ISIC2016 (heart and lungs segmentation on chest X-ray images and skin lesion segmentation on dermoscopy images). We have achieved competitive results using the same amount of training data compared to the state-of-the-art methods. More importantly, we demonstrate that our framework is extremely data-efficient, and it can achieve reliable results using extremely limited training data. Furthermore, the proposed method is rotationally invariant and insensitive to initialization.

Minimum-Data-Driven Guidance for Impact Angle Control

This paper investigates the impact-angle-control guidance problem for varying-speed flight vehicles with constrained acceleration. A learning-based bias proportional navigation guidance (L-BPN) law is proposed to achieve impact-angle-constrained impact by constructing a deep neural network (DNN) for nonlinear mapping between the impact angle and the bias term. During the process of dataset establishment, the impact of state variables is evaluated by sensitivity analysis to minimize the quantity of training data. This approach also effectively accelerates sample generation and improves the training efficiency. The simulation results verify the effectiveness of the proposed L-BPN law and demonstrate its advantages over the existing algorithms.

Improving the accuracy of blood pressure estimation using a photoplethysmogram due to additional training of an artificial neural network

The paper presents a neural network algorithm for cuffless blood pressure measurement based on computer analysis of the photoplethysmogram signal. An approach is proposed that allows you to significantly reduce the requirements for the size of the training sample required to build an effective machine learning model. The essence of the approach is the procedure for retraining a pre-trained neural network model using a limited amount of training data. At the same time, it is possible to achieve such values of the accuracy of estimating blood pressure values as in full-fledged training, but at the same time, the practical use of the method is significantly simplified.

Depth estimation and 3D reconstruction from UAV-borne imagery: Evaluation on the UseGeo dataset

Depth estimation and 3D model reconstruction from aerial imagery is an important task in photogrammetry, remote sensing, and computer vision. To compare the performance of different image-based approaches, this study presents a benchmark for UAV-based aerial imagery using the UseGeo dataset. The contributions include the release of various evaluation routines on GitHub, as well as a comprehensive comparison of baseline approaches, such as methods for offline multi-view 3D reconstruction resulting in point clouds and triangle meshes, online multi-view depth estimation, as well as single-image depth estimation using self-supervised deep learning. With the release of our evaluation routines, we aim to provide a universal protocol for the evaluation of depth estimation and 3D reconstruction methods on the UseGeo dataset. The conducted experiments and analyses show that each method excels in a different category: the depth estimation from COLMAP outperforms that of the other approaches, ACMMP achieves the lowest error and highest completeness for point clouds, while OpenMVS produces triangle meshes with the lowest error. Among the online methods for depth estimation, the approach from the Plane-Sweep Library outperforms the FaSS-MVS approach, while the latter achieves the lowest processing time. And even though the particularly challenging nature of the dataset and the small amount of training data leads to a significantly higher error in the results of the self-supervised single-image depth estimation approach, it outperforms all other approaches in terms of processing time and frame rate. In our evaluation, we have also considered modern learning-based approaches that can be used for image-based 3D reconstruction, such as NeRFs. However, due to the significantly lower quality of the resulting 3D models, we have only included a qualitative comparison between NeRF-based and conventional approaches in the scope of this work.