Much of the world's most valued data is stored in relational databases and data warehouses, where the data is organized into many tables connected by primary-foreign key relations. However, building machine learning models using this data is both challenging and time consuming. The core problem is that no machine learning method is capable of learning on multiple tables interconnected by primary-foreign key relations. Current methods can only learn from a single table, so the data must first be manually joined and aggregated into a single training table, the process known as feature engineering. Feature engineering is slow, error prone and leads to suboptimal models. Here we introduce an end-to-end deep representation learning approach to directly learn on data laid out across multiple tables. We name our approach Relational Deep Learning (RDL). The core idea is to view relational databases as a temporal, heterogeneous graph, with a node for each row in each table, and edges specified by primary-foreign key links. Message Passing Graph Neural Networks can then automatically learn across the graph to extract representations that leverage all input data, without any manual feature engineering. Relational Deep Learning leads to more accurate models that can be built much faster. To facilitate research in this area, we develop RelBench, a set of benchmark datasets and an implementation of Relational Deep Learning. The data covers a wide spectrum, from discussions on Stack Exchange to book reviews on the Amazon Product Catalog. Overall, we define a new research area that generalizes graph machine learning and broadens its applicability to a wide set of AI use cases.
Deep learning (DL) jobs use multi-dimensional parallelism, i.e. combining data, model, and pipeline parallelism, to use large GPU clusters efficiently. Long-running jobs may experience changes to their GPU allocation: (i) resource elasticity during training adds or removes GPUs; (ii) hardware maintenance may require redeployment on different GPUs; and (iii) GPU failures force jobs to run with fewer devices. Current DL frameworks tie jobs to a set of GPUs and thus lack support for these scenarios. In particular, they cannot change the multi-dimensional parallelism of an already-running job in an efficient and model-independent way. We describe Scalai, a state management library for DL systems that enables jobs to change their parallelism dynamically after the GPU allocation is updated at runtime. Scalai achieves this through a new abstraction, a parallelizable tensor collection (PTC), that externalizes the job state during training. After a GPU change, Scalai uses the PTC to transform the job state: the PTC repartitions the dataset state under data parallelism and exposes it to DL workers through a virtual file system; and the PTC obtains the model state as partitioned checkpoints and transforms them to reflect the new parallelization configuration. For efficiency, Scalai executes PTC transformations in parallel with minimum data movement between workers. Our experiments show that Scalai enables DL jobs to support dynamic parallelization with low overhead.
During recent years, there here has been a boom in terms of deep learning use for handwriting analysis and recognition. One main application for handwriting analysis is early detection and diagnosis in the health field. Unfortunately, most real case problems still suffer a scarcity of data, which makes difficult the use of deep learning-based models. To alleviate this problem, some works resort to synthetic data generation. Lately, more works are directed towards guided data synthetic generation, a generation that uses the domain and data knowledge to generate realistic data that can be useful to train deep learning models. In this work, we combine the domain knowledge about the Alzheimer's disease for handwriting and use it for a more guided data generation. Concretely, we have explored the use of in-air movements for synthetic data generation.
Table recognition is using the computer to automatically understand the table, to detect the position of the table from the document or picture, and to correctly extract and identify the internal structure and content of the table. After earlier mainstream approaches based on heuristic rules and machine learning, the development of deep learning techniques has brought a new paradigm to this field. This review mainly discusses the table recognition problem from five aspects. The first part introduces data sets, benchmarks, and commonly used evaluation indicators. This section selects representative data sets, benchmarks, and evaluation indicators that are frequently used by researchers. The second part introduces the table recognition model. This survey introduces the development of the table recognition model, especially the table recognition model based on deep learning. It is generally accepted that table recognition is divided into two stages: table detection and table structure recognition. This section introduces the models that follow this paradigm (TD and TSR). The third part is the End-to-End method, this section introduces some scholars' attempts to use an end-to-end approach to solve the table recognition problem once and for all and the part are Data-centric methods, such as data augmentation, aligning benchmarks, and other methods. The fourth part is the data-centric approach, such as data enhancement, alignment benchmark, and so on. The fifth part summarizes and compares the experimental data in the field of form recognition, and analyzes the mainstream and more advantageous methods. Finally, this paper also discusses the possible development direction and trend of form processing in the future, to provide some ideas for researchers in the field of table recognition. (Resource will be released at https://github.com/Wa1den-jy/Topic-on-Table-Recognition .)
We present a deep learning framework for volumetric speckle reduction in optical coherence tomography (OCT) based on a conditional generative adversarial network (cGAN) that leverages the volumetric nature of OCT data. In order to utilize the volumetric nature of OCT data, our network takes partial OCT volumes as input, resulting in artifact-free despeckled volumes that exhibit excellent speckle reduction and resolution preservation in all three dimensions. Furthermore, we address the ongoing challenge of generating ground truth data for supervised speckle suppression deep learning frameworks by using volumetric non-local means despeckling-TNode to generate training data. We show that, while TNode processing is computationally demanding, it serves as a convenient, accessible gold-standard source for training data; our cGAN replicates efficient suppression of speckle while preserving tissue structures with dimensions approaching the system resolution of non-local means despeckling while being two orders of magnitude faster than TNode. We demonstrate fast, effective, and high-quality despeckling of the proposed network in different tissue types acquired with three different OCT systems compared to existing deep learning methods. The open-source nature of our work facilitates re-training and deployment in any OCT system with an all-software implementation, working around the challenge of generating high-quality, speckle-free training data.
In machine learning energy potentials for atomic systems, forces are commonly obtained as the negative derivative of the energy function with respect to atomic positions. To quantify aleatoric uncertainty in the predicted energies, a widely used modeling approach involves predicting both a mean and variance for each energy value. However, this model is not differentiable under the usual white noise assumption, so energy uncertainty does not naturally translate to force uncertainty. In this work we propose a machine learning potential energy model in which energy and force aleatoric uncertainty are linked through a spatially correlated noise process. We demonstrate our approach on an equivariant messages passing neural network potential trained on energies and forces on two out-of-equilibrium molecular datasets. Furthermore, we also show how to obtain epistemic uncertainties in this setting based on a Bayesian interpretation of deep ensemble models.
Hyperparameters of Deep Learning (DL) pipelines are crucial for their downstream performance. While a large number of methods for Hyperparameter Optimization (HPO) have been developed, their incurred costs are often untenable for modern DL. Consequently, manual experimentation is still the most prevalent approach to optimize hyperparameters, relying on the researcher's intuition, domain knowledge, and cheap preliminary explorations. To resolve this misalignment between HPO algorithms and DL researchers, we propose PriorBand, an HPO algorithm tailored to DL, able to utilize both expert beliefs and cheap proxy tasks. Empirically, we demonstrate PriorBand's efficiency across a range of DL benchmarks and show its gains under informative expert input and robustness against poor expert beliefs
The modeling and forecasting of sea ice conditions in the Arctic region are important tasks for ship routing, offshore oil production, and environmental monitoring. We propose the adaptive surrogate modeling approach named LANE-SI (Lightweight Automated Neural Ensembling for Sea Ice) that uses ensemble of relatively simple deep learning models with different loss functions for forecasting of spatial distribution for sea ice concentration in the specified water area. Experimental studies confirm the quality of a long-term forecast based on a deep learning model fitted to the specific water area is comparable to resource-intensive physical modeling, and for some periods of the year, it is superior. We achieved a 20% improvement against the state-of-the-art physics-based forecast system SEAS5 for the Kara Sea.
Inferring parameters of macro-kinetic growth models, typically represented by Ordinary Differential Equations (ODE), from the experimental data is a crucial step in bioprocess engineering. Conventionally, estimates of the parameters are obtained by fitting the mechanistic model to observations. Fitting, however, requires a significant computational power. Specifically, during the development of new bioprocesses that use previously unknown organisms or strains, efficient, robust, and computationally cheap methods for parameter estimation are of great value. In this work, we propose using Deep Neural Networks (NN) for directly predicting parameters of mechanistic models given observations. The approach requires spending computational resources for training a NN, nonetheless, once trained, such a network can provide parameter estimates orders of magnitude faster than conventional methods. We consider a training procedure that combines Neural Networks and mechanistic models. We demonstrate the performance of the proposed algorithms on data sampled from several mechanistic models used in bioengineering describing a typical industrial batch process and compare the proposed method, a typical gradient-based fitting procedure, and the combination of the two. We find that, while Neural Network estimates are slightly improved by further fitting, these estimates are measurably better than the fitting procedure alone.
Protein arginine methylation (PRme), as one post-translational modification, plays a critical role in numerous cellular processes and regulates critical cellular functions. Though several in silico models for predicting PRme sites have been reported, new models may be required to develop due to the significant increase of identified PRme sites. In this study, we constructed multiple machine-learning and deep-learning models. The deep-learning model CNN combined with the One-Hot coding showed the best performance, dubbed CNNArginineMe. CNNArginineMe performed best in AUC scoring metrics in comparisons with several reported predictors. Additionally, we employed CNNArginineMe to predict arginine methylation proteome and performed functional analysis. The arginine methylated proteome is significantly enriched in the amyotrophic lateral sclerosis (ALS) pathway. CNNArginineMe is freely available at https://github.com/guoyangzou/CNNArginineMe.
The challenge of teaching robots to perform dexterous manipulation, dynamic locomotion, or whole--body manipulation from a small number of demonstrations is an important research field that has attracted interest from across the robotics community. In this work, we propose a novel approach by joining the theories of Koopman Operators and Dynamic Movement Primitives to Learning from Demonstration. Our approach, named \gls{admd}, projects nonlinear dynamical systems into linear latent spaces such that a solution reproduces the desired complex motion. Use of an autoencoder in our approach enables generalizability and scalability, while the constraint to a linear system attains interpretability. Our results are comparable to the Extended Dynamic Mode Decomposition on the LASA Handwriting dataset but with training on only a small fractions of the letters.
Deep learning has yielded remarkable outcomes in various domains. However, the challenge of requiring large-scale labeled samples still persists in deep learning. Thus, data augmentation has been introduced as a critical strategy to train deep learning models. However, data augmentation suffers from information loss and poor performance in small sample environments. To overcome these drawbacks, we propose a feature augmentation method based on shape space theory, i.e., feature augmentation on Geodesic curve, called FAGC in brevity.First, we extract features from the image with the neural network model. Then, the multiple image features are projected into a pre-shape space as features. In the pre-shape space, a Geodesic curve is built to fit the features. Finally, the many generated features on the Geodesic curve are used to train the various machine learning models. The FAGC module can be seamlessly integrated with most machine learning methods. And the proposed method is simple, effective and insensitive for the small sample datasets.Several examples demonstrate that the FAGC method can greatly improve the performance of the data preprocessing model in a small sample environment.
Deep learning based intrusion detection systems (DL-based IDS) have emerged as one of the best choices for providing security solutions against various network intrusion attacks. However, due to the emergence and development of adversarial deep learning technologies, it becomes challenging for the adoption of DL models into IDS. In this paper, we propose a novel IDS architecture that can enhance the robustness of IDS against adversarial attacks by combining conventional machine learning (ML) models and Deep Learning models. The proposed DLL-IDS consists of three components: DL-based IDS, adversarial example (AE) detector, and ML-based IDS. We first develop a novel AE detector based on the local intrinsic dimensionality (LID). Then, we exploit the low attack transferability between DL models and ML models to find a robust ML model that can assist us in determining the maliciousness of AEs. If the input traffic is detected as an AE, the ML-based IDS will predict the maliciousness of input traffic, otherwise the DL-based IDS will work for the prediction. The fusion mechanism can leverage the high prediction accuracy of DL models and low attack transferability between DL models and ML models to improve the robustness of the whole system. In our experiments, we observe a significant improvement in the prediction performance of the IDS when subjected to adversarial attack, achieving high accuracy with low resource consumption.
With the advance of the powerful heterogeneous, parallel and distributed computing systems and ever increasing immense amount of data, machine learning has become an indispensable part of cutting-edge technology, scientific research and consumer products. In this study, we present a review of modern machine and deep learning. We provide a high-level overview for the latest advanced machine learning algorithms, applications, and frameworks. Our discussion encompasses parallel distributed learning, deep learning as well as federated learning. As a result, our work serves as an introductory text to the vast field of modern machine learning.
At the heart of foundation models is the philosophy of "more is different", exemplified by the astonishing success in computer vision and natural language processing. However, the challenges of optimization and inherent complexity of transformer models call for a paradigm shift towards simplicity. In this study, we introduce VanillaNet, a neural network architecture that embraces elegance in design. By avoiding high depth, shortcuts, and intricate operations like self-attention, VanillaNet is refreshingly concise yet remarkably powerful. Each layer is carefully crafted to be compact and straightforward, with nonlinear activation functions pruned after training to restore the original architecture. VanillaNet overcomes the challenges of inherent complexity, making it ideal for resource-constrained environments. Its easy-to-understand and highly simplified architecture opens new possibilities for efficient deployment. Extensive experimentation demonstrates that VanillaNet delivers performance on par with renowned deep neural networks and vision transformers, showcasing the power of minimalism in deep learning. This visionary journey of VanillaNet has significant potential to redefine the landscape and challenge the status quo of foundation model, setting a new path for elegant and effective model design. Pre-trained models and codes are available at https://github.com/huawei-noah/VanillaNet and https://gitee.com/mindspore/models/tree/master/research/cv/vanillanet.
Deep learning datasets are expanding at an unprecedented pace, creating new challenges for data processing in model training pipelines. A crucial aspect of these pipelines is dataset shuffling, which significantly improves unbiased learning and convergence accuracy by adhering to the principles of random sampling. However, loading shuffled data for large datasets incurs significant overhead in the deep learning pipeline and severely impacts the end-to-end training throughput. To mitigate this, current deep learning systems often resort to partial dataset shuffling, sacrificing global randomness to maintain acceptable training throughput on large datasets, still leaving global shuffling efficiency issues not fully explored. In this work, we present RINAS, a data loading framework that systematically addresses the performance bottleneck of loading global shuffled datasets. Our key contribution is to offer an intra-batch unordered data fetching approach, which unleashes unexplored parallelism of data loading. We implement RINAS under the PyTorch framework for common dataset libraries HuggingFace and TorchVision. Our experimental results show that RINAS improves the throughput of general language model training and vision model training by up to 59% and 89%, respectively.
[Objective] This study aims to introduce a novel back projection-induced U-Net-shaped architecture, called ReconU-Net, for deep learning-based direct positron emission tomography (PET) image reconstruction. Additionally, our objective is to analyze the behavior of direct PET image reconstruction and gain deeper insights by comparing the proposed ReconU-Net architecture with other encoder-decoder architectures without skip connections. [Approach] The proposed ReconU-Net architecture uniquely integrates the physical model of the back projection operation into the skip connection. This distinctive feature facilitates the effective transfer of intrinsic spatial information from the input sinogram to the reconstructed image via an embedded physical model. The proposed ReconU-Net was trained using Monte Carlo simulation data from the Brainweb phantom and tested on both simulated and real Hoffman brain phantom data. [Main results] The proposed ReconU-Net method generated a reconstructed image with a more accurate structure compared to other deep learning-based direct reconstruction methods. Further analysis showed that the proposed ReconU-Net architecture has the ability to transfer features of multiple resolutions, especially non-abstract high-resolution information, through skip connections. Despite limited training on simulated data, the proposed ReconU-Net successfully reconstructed the real Hoffman brain phantom, unlike other deep learning-based direct reconstruction methods, which failed to produce a reconstructed image. [Significance] The proposed ReconU-Net can improve the fidelity of direct PET image reconstruction, even when dealing with small training datasets, by leveraging the synergistic relationship between data-driven modeling and the physics model of the imaging process.
We present the first deep learning model for segmenting galactic spiral arms and bars. In a blinded assessment by expert astronomers, our predicted spiral arm masks are preferred over both current automated methods (99% of evaluations) and our original volunteer labels (79% of evaluations). Experts rated our spiral arm masks as `mostly good' to `perfect' in 89% of evaluations. Bar lengths trivially derived from our predicted bar masks are in excellent agreement with a dedicated crowdsourcing project. The pixelwise precision of our masks, previously impossible at scale, will underpin new research into how spiral arms and bars evolve.
Real-world datasets are often highly class-imbalanced, which can adversely impact the performance of deep learning models. The majority of research on training neural networks under class imbalance has focused on specialized loss functions, sampling techniques, or two-stage training procedures. Notably, we demonstrate that simply tuning existing components of standard deep learning pipelines, such as the batch size, data augmentation, optimizer, and label smoothing, can achieve state-of-the-art performance without any such specialized class imbalance methods. We also provide key prescriptions and considerations for training under class imbalance, and an understanding of why imbalance methods succeed or fail.
We propose a classifier that can identify ten common home network problems based on the raw textual output of networking tools such as ping, dig, and ip. Our deep learning model uses an encoder-only transformer architecture with a particular pre-tokenizer that we propose for splitting the tool output into token sequences. The use of transformers distinguishes our approach from related work on network problem classification, which still primarily relies on non-deep-learning methods. Our model achieves high accuracy in our experiments, demonstrating the high potential of transformer-based problem classification for the home network.