PyTorchKR Papers

최신 논문

총 10,240편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2304.08502

CyFormer: Accurate State-of-Health Prediction of Lithium-Ion Batteries via Cyclic Attention

Predicting the State-of-Health (SoH) of lithium-ion batteries is a fundamental task of battery management systems on electric vehicles. It aims at estimating future SoH based on historical aging data. Most existing deep learning methods rely on filter-based feature extractors (e.g., CNN or Kalman filters) and recurrent time sequence models. Though efficient, they generally ignore cyclic features and the domain gap between training and testing batteries. To address this problem, we present CyFormer, a transformer-based cyclic time sequence model for SoH prediction. Instead of the conventional CNN-RNN structure, we adopt an encoder-decoder architecture. In the encoder, row-wise and column-wise attention blocks effectively capture intra-cycle and inter-cycle connections and extract cyclic features. In the decoder, the SoH queries cross-attend to these features to form the final predictions. We further utilize a transfer learning strategy to narrow the domain gap between the training and testing set. To be specific, we use fine-tuning to shift the model to a target working condition. Finally, we made our model more efficient by pruning. The experiment shows that our method attains an MAE of 0.75\% with only 10\% data for fine-tuning on a testing battery, surpassing prior methods by a large margin. Effective and robust, our method provides a potential solution for all cyclic time sequence prediction tasks.

arXiv:2304.07862

PBNR: Prompt-based News Recommender System

Online news platforms often use personalized news recommendation methods to help users discover articles that align with their interests. These methods typically predict a matching score between a user and a candidate article to reflect the user's preference for the article. Some previous works have used language model techniques, such as the attention mechanism, to capture users' interests based on their past behaviors, and to understand the content of articles. However, these existing model architectures require adjustments if additional information is taken into account. Pre-trained large language models, which can better capture word relationships and comprehend contexts, have seen a significant development in recent years, and these pre-trained models have the advantages of transfer learning and reducing the training time for downstream tasks. Meanwhile, prompt learning is a newly developed technique that leverages pre-trained language models by building task-specific guidance for output generations. To leverage textual information in news articles, this paper introduces the pre-trained large language model and prompt-learning to the community of news recommendation. The proposed model "prompt-based news recommendation" (PBNR) treats the personalized news recommendation as a text-to-text language task and designs personalized prompts to adapt to the pre-trained language model -- text-to-text transfer transformer (T5). Experimental studies using the Microsoft News dataset show that PBNR is capable of making accurate recommendations by taking into account various lengths of past behaviors of different users. PBNR can also easily adapt to new information without changing the model architecture and the training objective. Additionally, PBNR can make recommendations based on users' specific requirements, allowing human-computer interaction in the news recommendation field.

arXiv:2304.07469

Generating an interactive online map of future sea level rise along the North Shore of Vancouver: methods and insights on enabling geovisualisation for coastal communities

Contemporary sea level rise (SLR) research seldom considers enabling effective geovisualisation for the communities. This lack of knowledge transfer impedes raising awareness on climate change and its impacts. The goal of this study is to produce an online SLR map accessible to the public that allows them to interact with evolving high-resolution geospatial data and techniques. The study area was the North Shore of Vancouver, British Columbia, Canada. While typically coarser resolution (10m+/pixel) Digital Elevation Models have been used by previous studies, we explored an open access airborne 1 metre LiDAR which has a higher resolution and vertical accuracy and can penetrate tree cover at a higher degree than most satellite imagery. A bathtub method model with hydrologic connectivity was used to delineate the inundation zones for various SLR scenarios which allows for a not overly complex model and process using standard tools such as ArcGIS and QGIS with similar levels of accuracy as more complex models, especially with the high-resolution data. Deep Learning and 3D visualizations were used to create past, present, and modelled future Land Use/Land Cover and 3D flyovers. Analysis of the possible impacts of 1m, 2m, 3m, and 4m SLR over the unique coastline, terrain and land use was detailed. The generated interactive online map helps local communities visualise and understand the future of their coastlines. We have provided a detailed methodology and the methods and results are easily reproducible for other regions. Such initiatives can help popularise community-focused geovisualisation to raise awareness about SLR.

arXiv:2304.07967

360$^\circ$ High-Resolution Depth Estimation via Uncertainty-aware Structural Knowledge Transfer

To predict high-resolution (HR) omnidirectional depth map, existing methods typically leverage HR omnidirectional image (ODI) as the input via fully-supervised learning. However, in practice, taking HR ODI as input is undesired due to resource-constrained devices. In addition, depth maps are often with lower resolution than color images. Therefore, in this paper, we explore for the first time to estimate the HR omnidirectional depth directly from a low-resolution (LR) ODI, when no HR depth GT map is available. Our key idea is to transfer the scene structural knowledge from the HR image modality and the corresponding LR depth maps to achieve the goal of HR depth estimation without any extra inference cost. Specifically, we introduce ODI super-resolution (SR) as an auxiliary task and train both tasks collaboratively in a weakly supervised manner to boost the performance of HR depth estimation. The ODI SR task extracts the scene structural knowledge via uncertainty estimation. Buttressed by this, a scene structural knowledge transfer (SSKT) module is proposed with two key components. First, we employ a cylindrical implicit interpolation function (CIIF) to learn cylindrical neural interpolation weights for feature up-sampling and share the parameters of CIIFs between the two tasks. Then, we propose a feature distillation (FD) loss that provides extra structural regularization to help the HR depth estimation task learn more scene structural knowledge. Extensive experiments demonstrate that our weakly-supervised method outperforms baseline methods, and even achieves comparable performance with the fully-supervised methods.

arXiv:2304.07574

Exploring Incompatible Knowledge Transfer in Few-shot Image Generation

Few-shot image generation (FSIG) learns to generate diverse and high-fidelity images from a target domain using a few (e.g., 10) reference samples. Existing FSIG methods select, preserve and transfer prior knowledge from a source generator (pretrained on a related domain) to learn the target generator. In this work, we investigate an underexplored issue in FSIG, dubbed as incompatible knowledge transfer, which would significantly degrade the realisticness of synthetic samples. Empirical observations show that the issue stems from the least significant filters from the source generator. To this end, we propose knowledge truncation to mitigate this issue in FSIG, which is a complementary operation to knowledge preservation and is implemented by a lightweight pruning-based method. Extensive experiments show that knowledge truncation is simple and effective, consistently achieving state-of-the-art performance, including challenging setups where the source and target domains are more distant. Project Page: yunqing-me.github.io/RICK.

arXiv:2304.07869공식 코드 · Papers with Code

Neural Machine Translation For Low Resource Languages

Neural Machine translation is a challenging task due to the inherent complex nature and the fluidity that natural languages bring. Nonetheless, in recent years, it has achieved state-of-the-art performance in several language pairs. Although, a lot of traction can be seen in the areas of multilingual neural machine translation (MNMT) in the recent years, there are no comprehensive survey done to identify what approaches work well. The goal of this paper is to investigate the realm of low resource languages and build a Neural Machine Translation model to achieve state-of-the-art results. The paper looks to build upon the mBART language model and explore strategies to augment it with various NLP and Deep Learning techniques like back translation and transfer learning. This implementation tries to unpack the architecture of the NMT application and determine the different components which offers us opportunities to amend the said application within the purview of the low resource languages problem space.

arXiv:2304.07759공식 코드 · Papers with Code

MisRoBÆRTa: Transformers versus Misinformation

Misinformation is considered a threat to our democratic values and principles. The spread of such content on social media polarizes society and undermines public discourse by distorting public perceptions and generating social unrest while lacking the rigor of traditional journalism. Transformers and transfer learning proved to be state-of-the-art methods for multiple well-known natural language processing tasks. In this paper, we propose MisRoB{\AE}RTa, a novel transformer-based deep neural ensemble architecture for misinformation detection. MisRoB{\AE}RTa takes advantage of two transformers (BART \& RoBERTa) to improve the classification performance. We also benchmarked and evaluated the performances of multiple transformers on the task of misinformation detection. For training and testing, we used a large real-world news articles dataset labeled with 10 classes, addressing two shortcomings in the current research: increasing the size of the dataset from small to large, and moving the focus of fake news detection from binary classification to multi-class classification. For this dataset, we manually verified the content of the news articles to ensure that they were correctly labeled. The experimental results show that the accuracy of transformers on the misinformation detection problem was significantly influenced by the method employed to learn the context, dataset size, and vocabulary dimension. We observe empirically that the best accuracy performance among the classification models that use only one transformer is obtained by BART, while DistilRoBERTa obtains the best accuracy in the least amount of time required for fine-tuning and training. The proposed MisRoB{\AE}RTa outperforms the other transformer models in the task of misinformation detection. To arrive at this conclusion, we performed ample ablation and sensitivity testing with MisRoB{\AE}RTa on two datasets.

arXiv:2304.07453

Context-aware Domain Adaptation for Time Series Anomaly Detection

Time series anomaly detection is a challenging task with a wide range of real-world applications. Due to label sparsity, training a deep anomaly detector often relies on unsupervised approaches. Recent efforts have been devoted to time series domain adaptation to leverage knowledge from similar domains. However, existing solutions may suffer from negative knowledge transfer on anomalies due to their diversity and sparsity. Motivated by the empirical study of context alignment between two domains, we aim to transfer knowledge between two domains via adaptively sampling context information for two domains. This is challenging because it requires simultaneously modeling the complex in-domain temporal dependencies and cross-domain correlations while exploiting label information from the source domain. To this end, we propose a framework that combines context sampling and anomaly detection into a joint learning procedure. We formulate context sampling into the Markov decision process and exploit deep reinforcement learning to optimize the time series domain adaptation process via context sampling and design a tailored reward function to generate domain-invariant features that better align two domains for anomaly detection. Experiments on three public datasets show promise for knowledge transfer between two similar domains and two entirely different domains.

arXiv:2304.07017

Symbiotic Message Passing Model for Transfer Learning between Anti-Fungal and Anti-Bacterial Domains

Machine learning, and representation learning in particular, has the potential to facilitate drug discovery by screening billions of compounds. For example, a successful approach is representing the molecules as a graph and utilizing graph neural networks (GNN). Yet, these approaches still require experimental measurements of thousands of compounds to construct a proper training set. While in some domains it is easier to acquire experimental data, in others it might be more limited. For example, it is easier to test the compounds on bacteria than perform in-vivo experiments. Thus, a key question is how to utilize information from a large available dataset together with a small subset of compounds where both domains are measured to predict compounds' effect on the second, experimentally less available domain. Current transfer learning approaches for drug discovery, including training of pre-trained modules or meta-learning, have limited success. In this work, we develop a novel method, named Symbiotic Message Passing Neural Network (SMPNN), for merging graph-neural-network models from different domains. Using routing new message passing lanes between them, our approach resolves some of the potential conflicts between the different domains, and implicit constraints induced by the larger datasets. By collecting public data and performing additional high-throughput experiments, we demonstrate the advantage of our approach by predicting anti-fungal activity from anti-bacterial activity. We compare our method to the standard transfer learning approach and show that SMPNN provided better and less variable performances. Our approach is general and can be used to facilitate information transfer between any two domains such as different organisms, different organelles, or different environments.

arXiv:2304.06972

Multi-fidelity prediction of fluid flow and temperature field based on transfer learning using Fourier Neural Operator

Data-driven prediction of fluid flow and temperature distribution in marine and aerospace engineering has received extensive research and demonstrated its potential in real-time prediction recently. However, usually large amounts of high-fidelity data are required to describe and accurately predict the complex physical information, while in reality, only limited high-fidelity data is available due to the high experiment/computational cost. Therefore, this work proposes a novel multi-fidelity learning method based on the Fourier Neural Operator by jointing abundant low-fidelity data and limited high-fidelity data under transfer learning paradigm. First, as a resolution-invariant operator, the Fourier Neural Operator is first and gainfully applied to integrate multi-fidelity data directly, which can utilize the scarce high-fidelity data and abundant low-fidelity data simultaneously. Then, the transfer learning framework is developed for the current task by extracting the rich low-fidelity data knowledge to assist high-fidelity modeling training, to further improve data-driven prediction accuracy. Finally, three typical fluid and temperature prediction problems are chosen to validate the accuracy of the proposed multi-fidelity model. The results demonstrate that our proposed method has high effectiveness when compared with other high-fidelity models, and has the high modeling accuracy of 99% for all the selected physical field problems. Significantly, the proposed multi-fidelity learning method has the potential of a simple structure with high precision, which can provide a reference for the construction of the subsequent model.

arXiv:2304.06957

MVP-SEG: Multi-View Prompt Learning for Open-Vocabulary Semantic Segmentation

CLIP (Contrastive Language-Image Pretraining) is well-developed for open-vocabulary zero-shot image-level recognition, while its applications in pixel-level tasks are less investigated, where most efforts directly adopt CLIP features without deliberative adaptations. In this work, we first demonstrate the necessity of image-pixel CLIP feature adaption, then provide Multi-View Prompt learning (MVP-SEG) as an effective solution to achieve image-pixel adaptation and to solve open-vocabulary semantic segmentation. Concretely, MVP-SEG deliberately learns multiple prompts trained by our Orthogonal Constraint Loss (OCLoss), by which each prompt is supervised to exploit CLIP feature on different object parts, and collaborative segmentation masks generated by all prompts promote better segmentation. Moreover, MVP-SEG introduces Global Prompt Refining (GPR) to further eliminate class-wise segmentation noise. Experiments show that the multi-view prompts learned from seen categories have strong generalization to unseen categories, and MVP-SEG+ which combines the knowledge transfer stage significantly outperforms previous methods on several benchmarks. Moreover, qualitative results justify that MVP-SEG does lead to better focus on different local parts.

arXiv:2304.06495

An embedding for EEG signals learned using a triplet loss

Neurophysiological time series recordings like the electroencephalogram (EEG) or local field potentials are obtained from multiple sensors. They can be decoded by machine learning models in order to estimate the ongoing brain state of a patient or healthy user. In a brain-computer interface (BCI), this decoded brain state information can be used with minimal time delay to either control an application, e.g., for communication or for rehabilitation after stroke, or to passively monitor the ongoing brain state of the subject, e.g., in a demanding work environment. A specific challenge in such decoding tasks is posed by the small dataset sizes in BCI compared to other domains of machine learning like computer vision or natural language processing. A possibility to tackle classification or regression problems in BCI despite small training data sets is through transfer learning, which utilizes data from other sessions, subjects or even datasets to train a model. In this exploratory study, we propose novel domain-specific embeddings for neurophysiological data. Our approach is based on metric learning and builds upon the recently proposed ladder loss. Using embeddings allowed us to benefit, both from the good generalisation abilities and robustness of deep learning and from the fast training of classical machine learning models for subject-specific calibration. In offline analyses using EEG data of 14 subjects, we tested the embeddings' feasibility and compared their efficiency with state-of-the-art deep learning models and conventional machine learning pipelines. In summary, we propose the use of metric learning to obtain pre-trained embeddings of EEG-BCI data as a means to incorporate domain knowledge and to reach competitive performance on novel subjects with minimal calibration requirements.

arXiv:2304.06489

Domain Adaptation for Inertial Measurement Unit-based Human Activity Recognition: A Survey

Machine learning-based wearable human activity recognition (WHAR) models enable the development of various smart and connected community applications such as sleep pattern monitoring, medication reminders, cognitive health assessment, sports analytics, etc. However, the widespread adoption of these WHAR models is impeded by their degraded performance in the presence of data distribution heterogeneities caused by the sensor placement at different body positions, inherent biases and heterogeneities across devices, and personal and environmental diversities. Various traditional machine learning algorithms and transfer learning techniques have been proposed in the literature to address the underpinning challenges of handling such data heterogeneities. Domain adaptation is one such transfer learning techniques that has gained significant popularity in recent literature. In this paper, we survey the recent progress of domain adaptation techniques in the Inertial Measurement Unit (IMU)-based human activity recognition area, discuss potential future directions.

arXiv:2304.06033공식 코드 · Papers with Code

Quantifying the Impact of Data Characteristics on the Transferability of Sleep Stage Scoring Models

Deep learning models for scoring sleep stages based on single-channel EEG have been proposed as a promising method for remote sleep monitoring. However, applying these models to new datasets, particularly from wearable devices, raises two questions. First, when annotations on a target dataset are unavailable, which different data characteristics affect the sleep stage scoring performance the most and by how much? Second, when annotations are available, which dataset should be used as the source of transfer learning to optimize performance? In this paper, we propose a novel method for computationally quantifying the impact of different data characteristics on the transferability of deep learning models. Quantification is accomplished by training and evaluating two models with significant architectural differences, TinySleepNet and U-Time, under various transfer configurations in which the source and target datasets have different recording channels, recording environments, and subject conditions. For the first question, the environment had the highest impact on sleep stage scoring performance, with performance degrading by over 14% when sleep annotations were unavailable. For the second question, the most useful transfer sources for TinySleepNet and the U-Time models were MASS-SS1 and ISRUC-SG1, containing a high percentage of N1 (the rarest sleep stage) relative to the others. The frontal and central EEGs were preferred for TinySleepNet. The proposed approach enables full utilization of existing sleep datasets for training and planning model transfer to maximize the sleep stage scoring performance on a target problem when sleep annotations are limited or unavailable, supporting the realization of remote sleep monitoring.

arXiv:2304.06704

How Will It Drape Like? Capturing Fabric Mechanics from Depth Images

We propose a method to estimate the mechanical parameters of fabrics using a casual capture setup with a depth camera. Our approach enables to create mechanically-correct digital representations of real-world textile materials, which is a fundamental step for many interactive design and engineering applications. As opposed to existing capture methods, which typically require expensive setups, video sequences, or manual intervention, our solution can capture at scale, is agnostic to the optical appearance of the textile, and facilitates fabric arrangement by non-expert operators. To this end, we propose a sim-to-real strategy to train a learning-based framework that can take as input one or multiple images and outputs a full set of mechanical parameters. Thanks to carefully designed data augmentation and transfer learning protocols, our solution generalizes to real images despite being trained only on synthetic data, hence successfully closing the sim-to-real loop.Key in our work is to demonstrate that evaluating the regression accuracy based on the similarity at parameter space leads to an inaccurate distances that do not match the human perception. To overcome this, we propose a novel metric for fabric drape similarity that operates on the image domain instead on the parameter space, allowing us to evaluate our estimation within the context of a similarity rank. We show that out metric correlates with human judgments about the perception of drape similarity, and that our model predictions produce perceptually accurate results compared to the ground truth parameters.

arXiv:2304.06520

An Efficient Transfer Learning-based Approach for Apple Leaf Disease Classification

Correct identification and categorization of plant diseases are crucial for ensuring the safety of the global food supply and the overall financial success of stakeholders. In this regard, a wide range of solutions has been made available by introducing deep learning-based classification systems for different staple crops. Despite being one of the most important commercial crops in many parts of the globe, research proposing a smart solution for automatically classifying apple leaf diseases remains relatively unexplored. This study presents a technique for identifying apple leaf diseases based on transfer learning. The system extracts features using a pretrained EfficientNetV2S architecture and passes to a classifier block for effective prediction. The class imbalance issues are tackled by utilizing runtime data augmentation. The effect of various hyperparameters, such as input resolution, learning rate, number of epochs, etc., has been investigated carefully. The competence of the proposed pipeline has been evaluated on the apple leaf disease subset from the publicly available `PlantVillage' dataset, where it achieved an accuracy of 99.21%, outperforming the existing works.

arXiv:2304.06446

SpectFormer: Frequency and Attention is what you need in a Vision Transformer

Vision transformers have been applied successfully for image recognition tasks. There have been either multi-headed self-attention based (ViT \cite{dosovitskiy2020image}, DeIT, \cite{touvron2021training}) similar to the original work in textual models or more recently based on spectral layers (Fnet\cite{lee2021fnet}, GFNet\cite{rao2021global}, AFNO\cite{guibas2021efficient}). We hypothesize that both spectral and multi-headed attention plays a major role. We investigate this hypothesis through this work and observe that indeed combining spectral and multi-headed attention layers provides a better transformer architecture. We thus propose the novel Spectformer architecture for transformers that combines spectral and multi-headed attention layers. We believe that the resulting representation allows the transformer to capture the feature representation appropriately and it yields improved performance over other transformer representations. For instance, it improves the top-1 accuracy by 2\% on ImageNet compared to both GFNet-H and LiT. SpectFormer-S reaches 84.25\% top-1 accuracy on ImageNet-1K (state of the art for small version). Further, Spectformer-L achieves 85.7\% that is the state of the art for the comparable base version of the transformers. We further ensure that we obtain reasonable results in other scenarios such as transfer learning on standard datasets such as CIFAR-10, CIFAR-100, Oxford-IIIT-flower, and Standford Car datasets. We then investigate its use in downstream tasks such of object detection and instance segmentation on the MS-COCO dataset and observe that Spectformer shows consistent performance that is comparable to the best backbones and can be further optimized and improved. Hence, we believe that combined spectral and attention layers are what are needed for vision transformers.

arXiv:2304.05985

KubeEdge-Sedna v0.3: Towards Next-Generation Automatically Customized AI Engineering Scheme

The scale of the global edge AI market continues to grow. The current technical challenges that hinder the large-scale replication of edge AI are mainly small samples on the edge and heterogeneity of edge data. In addition, edge AI customers often have requirements for data security compliance and offline autonomy of edge AI services. Based on the lifelong learning method in the academic world, we formally define the problem of edge-cloud collaborative lifelong learning for the first time, and release the industry's first open-source edge-cloud collaborative lifelong learning. Edge-cloud collaborative lifelong learning adapts to data heterogeneity at different edge locations through (1) multi-task transfer learning to achieve accurate prediction of "thousands of people and thousands of faces"; (2) incremental processing of unknown tasks, the more systems learn and the smarter systems are with small samples, gradually realize AI engineering and automation; (3) Use the cloud-side knowledge base to remember new situational knowledge to avoid catastrophic forgetting; (4) The edge-cloud collaborative architecture enables data security compliance and edge AI services to be offline autonomy while applying cloud resources. This work hopes to help fundamentally solve the above-mentioned challenges of edge-cloud collaborative machine learning.

arXiv:2304.05871

Edge-cloud Collaborative Learning with Federated and Centralized Features

Federated learning (FL) is a popular way of edge computing that doesn't compromise users' privacy. Current FL paradigms assume that data only resides on the edge, while cloud servers only perform model averaging. However, in real-life situations such as recommender systems, the cloud server has the ability to store historical and interactive features. In this paper, our proposed Edge-Cloud Collaborative Knowledge Transfer Framework (ECCT) bridges the gap between the edge and cloud, enabling bi-directional knowledge transfer between both, sharing feature embeddings and prediction logits. ECCT consolidates various benefits, including enhancing personalization, enabling model heterogeneity, tolerating training asynchronization, and relieving communication burdens. Extensive experiments on public and industrial datasets demonstrate ECCT's effectiveness and potential for use in academia and industry.

arXiv:2304.05741

Learning to search for and detect objects in foveal images using deep learning

The human visual system processes images with varied degrees of resolution, with the fovea, a small portion of the retina, capturing the highest acuity region, which gradually declines toward the field of view's periphery. However, the majority of existing object localization methods rely on images acquired by image sensors with space-invariant resolution, ignoring biological attention mechanisms. As a region of interest pooling, this study employs a fixation prediction model that emulates human objective-guided attention of searching for a given class in an image. The foveated pictures at each fixation point are then classified to determine whether the target is present or absent in the scene. Throughout this two-stage pipeline method, we investigate the varying results obtained by utilizing high-level or panoptic features and provide a ground-truth label function for fixation sequences that is smoother, considering in a better way the spatial structure of the problem. Finally, we present a novel dual task model capable of performing fixation prediction and detection simultaneously, allowing knowledge transfer between the two tasks. We conclude that, due to the complementary nature of both tasks, the training process benefited from the sharing of knowledge, resulting in an improvement in performance when compared to the previous approach's baseline scores.