PyTorchKR Papers

최신 논문

총 10,240편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2304.03509

Local Rose Breeds Detection System Using Transfer Learning Techniques

Flower breed detection and giving details of that breed with the suggestion of cultivation processes and the way of taking care is important for flower cultivation, breed invention, and the flower business. Among all the local flowers in Bangladesh, the rose is one of the most popular and demanded flowers. Roses are the most desirable flower not only in Bangladesh but also throughout the world. Roses can be used for many other purposes apart from decoration. As roses have a great demand in the flower business so rose breed detection will be very essential. However, there is no remarkable work for breed detection of a particular flower unlike the classification of different flowers. In this research, we have proposed a model to detect rose breeds from images using transfer learning techniques. For such work in flowers, resources are not enough in image processing and classification, so we needed a large dataset of the massive number of images to train our model. we have used 1939 raw images of five different breeds and we have generated 9306 images for the training dataset and 388 images for the testing dataset to validate the model using augmentation. We have applied four transfer learning models in this research, which are Inception V3, ResNet50, Xception, and VGG16. Among these four models, VGG16 achieved the highest accuracy of 99%, which is an excellent outcome. Breed detection of a rose by using transfer learning methods is the first work on breed detection of a particular flower that is publicly available according to the study.

arXiv:2304.02877

Tag that issue: Applying API-domain labels in issue tracking systems

Labeling issues with the skills required to complete them can help contributors to choose tasks in Open Source Software projects. However, manually labeling issues is time-consuming and error-prone, and current automated approaches are mostly limited to classifying issues as bugs/non-bugs. We investigate the feasibility and relevance of automatically labeling issues with what we call "API-domains," which are high-level categories of APIs. Therefore, we posit that the APIs used in the source code affected by an issue can be a proxy for the type of skills (e.g., DB, security, UI) needed to work on the issue. We ran a user study (n=74) to assess API-domain labels' relevancy to potential contributors, leveraged the issues' descriptions and the project history to build prediction models, and validated the predictions with contributors (n=20) of the projects. Our results show that (i) newcomers to the project consider API-domain labels useful in choosing tasks, (ii) labels can be predicted with a precision of 84% and a recall of 78.6% on average, (iii) the results of the predictions reached up to 71.3% in precision and 52.5% in recall when training with a project and testing in another (transfer learning), and (iv) project contributors consider most of the predictions helpful in identifying needed skills. These findings suggest our approach can be applied in practice to automatically label issues, assisting developers in finding tasks that better match their skills.

arXiv:2304.02852

Classification of Skin Disease Using Transfer Learning in Convolutional Neural Networks

Automatic classification of skin disease plays an important role in healthcare especially in dermatology. Dermatologists can determine different skin diseases with the help of an android device and with the use of Artificial Intelligence. Deep learning requires a lot of time to train due to the number of sequential layers and input data involved. Powerful computer involving a Graphic Processing Unit is an ideal approach to the training process due to its parallel processing capability. This study gathered images of 7 types of skin disease prevalent in the Philippines for a skin disease classification system. There are 3400 images composed of different skin diseases like chicken pox, acne, eczema, Pityriasis rosea, psoriasis, Tinea corporis and vitiligo that was used for training and testing of different convolutional network models. This study used transfer learning to skin disease classification using pre-trained weights from different convolutional neural network models such as VGG16, VGG19, MobileNet, ResNet50, InceptionV3, Inception-ResNetV2, Xception, DenseNet121, DenseNet169, DenseNet201 and NASNet mobile. The MobileNet model achieved the highest accuracy, 94.1% and the VGG16 model achieved the lowest accuracy, 44.1%.

arXiv:2304.02993

Natural Language Robot Programming: NLP integrated with autonomous robotic grasping

In this paper, we present a grammar-based natural language framework for robot programming, specifically for pick-and-place tasks. Our approach uses a custom dictionary of action words, designed to store together words that share meaning, allowing for easy expansion of the vocabulary by adding more action words from a lexical database. We validate our Natural Language Robot Programming (NLRP) framework through simulation and real-world experimentation, using a Franka Panda robotic arm equipped with a calibrated camera-in-hand and a microphone. Participants were asked to complete a pick-and-place task using verbal commands, which were converted into text using Google's Speech-to-Text API and processed through the NLRP framework to obtain joint space trajectories for the robot. Our results indicate that our approach has a high system usability score. The framework's dictionary can be easily extended without relying on transfer learning or large data sets. In the future, we plan to compare the presented framework with different approaches of human-assisted pick-and-place tasks via a comprehensive user study.

arXiv:2304.02925

Computer-aided Diagnosis of Malaria through Transfer Learning using the ResNet50 Backbone

According to the World Malaria Report of 2022, 247 million cases of malaria and 619,000 related deaths were reported in 2021. This highlights the predominance of the disease, especially in the tropical and sub-tropical regions of Africa, parts of South-east Asia, Central and Southern America. Malaria is caused due to the Plasmodium parasite which is circulated through the bites of the female Anopheles mosquito. Hence, the detection of the parasite in human blood smears could confirm malarial infestation. Since the manual identification of Plasmodium is a lengthy and time-consuming task subject to variability in accuracy, we propose an automated, computer-aided diagnostic method to classify malarial thin smear blood cell images as parasitized and uninfected by using the ResNet50 Deep Neural Network. In this paper, we have used the pre-trained ResNet50 model on the open-access database provided by the National Library of Medicine's Lister Hill National Center for Biomedical Communication for 150 epochs. The results obtained showed accuracy, precision, and recall values of 98.75%, 99.3% and 99.5% on the ResNet50(proposed) model. We have compared these metrics with similar models such as VGG16, Watershed Segmentation and Random Forest, which showed better performance than traditional techniques as well.

arXiv:2304.02916

Efficient Audio Captioning Transformer with Patchout and Text Guidance

Automated audio captioning is multi-modal translation task that aim to generate textual descriptions for a given audio clip. In this paper we propose a full Transformer architecture that utilizes Patchout as proposed in [1], significantly reducing the computational complexity and avoiding overfitting. The caption generation is partly conditioned on textual AudioSet tags extracted by a pre-trained classification model which is fine-tuned to maximize the semantic similarity between AudioSet labels and ground truth captions. To mitigate the data scarcity problem of Automated Audio Captioning we introduce transfer learning from an upstream audio-related task and an enlarged in-domain dataset. Moreover, we propose a method to apply Mixup augmentation for AAC. Ablation studies are carried out to investigate how Patchout and text guidance contribute to the final performance. The results show that the proposed techniques improve the performance of our system and while reducing the computational complexity. Our proposed method received the Judges Award at the Task6A of DCASE Challenge 2022.

arXiv:2304.02798

Source-free Domain Adaptation Requires Penalized Diversity

While neural networks are capable of achieving human-like performance in many tasks such as image classification, the impressive performance of each model is limited to its own dataset. Source-free domain adaptation (SFDA) was introduced to address knowledge transfer between different domains in the absence of source data, thus, increasing data privacy. Diversity in representation space can be vital to a model`s adaptability in varied and difficult domains. In unsupervised SFDA, the diversity is limited to learning a single hypothesis on the source or learning multiple hypotheses with a shared feature extractor. Motivated by the improved predictive performance of ensembles, we propose a novel unsupervised SFDA algorithm that promotes representational diversity through the use of separate feature extractors with Distinct Backbone Architectures (DBA). Although diversity in feature space is increased, the unconstrained mutual information (MI) maximization may potentially introduce amplification of weak hypotheses. Thus we introduce the Weak Hypothesis Penalization (WHP) regularizer as a mitigation strategy. Our work proposes Penalized Diversity (PD) where the synergy of DBA and WHP is applied to unsupervised source-free domain adaptation for covariate shift. In addition, PD is augmented with a weighted MI maximization objective for label distribution shift. Empirical results on natural, synthetic, and medical domains demonstrate the effectiveness of PD under different distributional shifts.

arXiv:2304.02713

NUMSnet: Nested-U Multi-class Segmentation network for 3D Medical Image Stacks

Semantic segmentation for medical 3D image stacks enables accurate volumetric reconstructions, computer-aided diagnostics and follow up treatment planning. In this work, we present a novel variant of the Unet model called the NUMSnet that transmits pixel neighborhood features across scans through nested layers to achieve accurate multi-class semantic segmentations with minimal training data. We analyze the semantic segmentation performance of the NUMSnet model in comparison with several Unet model variants to segment 3-7 regions of interest using only 10% of images for training per Lung-CT and Heart-CT volumetric image stacks. The proposed NUMSnet model achieves up to 20% improvement in segmentation recall with 4-9% improvement in Dice scores for Lung-CT stacks and 2.5-10% improvement in Dice scores for Heart-CT stacks when compared to the Unet++ model. The NUMSnet model needs to be trained by ordered images around the central scan of each volumetric stack. Propagation of image feature information from the 6 nested layers of the Unet++ model are found to have better computation and segmentation performances than propagation of all up-sampling layers in a Unet++ model. The NUMSnet model achieves comparable segmentation performances to existing works, while being trained on as low as 5\% of the training images. Also, transfer learning allows faster convergence of the NUMSnet model for multi-class semantic segmentation from pathology in Lung-CT images to cardiac segmentations in Heart-CT stacks. Thus, the proposed model can standardize multi-class semantic segmentation on a variety of volumetric image stacks with minimal training dataset. This can significantly reduce the cost, time and inter-observer variabilities associated with computer-aided detections and treatment.

arXiv:2304.03193

Improving automatic endoscopic stone recognition using a multi-view fusion approach enhanced with two-step transfer learning

This contribution presents a deep-learning method for extracting and fusing image information acquired from different viewpoints, with the aim to produce more discriminant object features for the identification of the type of kidney stones seen in endoscopic images. The model was further improved with a two-step transfer learning approach and by attention blocks to refine the learned feature maps. Deep feature fusion strategies improved the results of single view extraction backbone models by more than 6% in terms of accuracy of the kidney stones classification.

arXiv:2304.02309

Multi-Domain Norm-referenced Encoding Enables Data Efficient Transfer Learning of Facial Expression Recognition

People can innately recognize human facial expressions in unnatural forms, such as when depicted on the unusual faces drawn in cartoons or when applied to an animal's features. However, current machine learning algorithms struggle with out-of-domain transfer in facial expression recognition (FER). We propose a biologically-inspired mechanism for such transfer learning, which is based on norm-referenced encoding, where patterns are encoded in terms of difference vectors relative to a domain-specific reference vector. By incorporating domain-specific reference frames, we demonstrate high data efficiency in transfer learning across multiple domains. Our proposed architecture provides an explanation for how the human brain might innately recognize facial expressions on varying head shapes (humans, monkeys, and cartoon avatars) without extensive training. Norm-referenced encoding also allows the intensity of the expression to be read out directly from neural unit activity, similar to face-selective neurons in the brain. Our model achieves a classification accuracy of 92.15\% on the FERG dataset with extreme data efficiency. We train our proposed mechanism with only 12 images, including a single image of each class (facial expression) and one image per domain (avatar). In comparison, the authors of the FERG dataset achieved a classification accuracy of 89.02\% with their FaceExpr model, which was trained on 43,000 images.

arXiv:2304.02263

Towards Efficient Task-Driven Model Reprogramming with Foundation Models

Vision foundation models exhibit impressive power, benefiting from the extremely large model capacity and broad training data. However, in practice, downstream scenarios may only support a small model due to the limited computational resources or efficiency considerations. Moreover, the data used for pretraining foundation models are usually invisible and very different from the target data of downstream tasks. This brings a critical challenge for the real-world application of foundation models: one has to transfer the knowledge of a foundation model to the downstream task that has a quite different architecture with only downstream target data. Existing transfer learning or knowledge distillation methods depend on either the same model structure or finetuning of the foundation model. Thus, naively introducing these methods can be either infeasible or very inefficient. To address this, we propose a Task-Driven Model Reprogramming (TDMR) framework. Specifically, we reprogram the foundation model to project the knowledge into a proxy space, which alleviates the adverse effect of task mismatch and domain inconsistency. Then, we reprogram the target model via progressive distillation from the proxy space to efficiently learn the knowledge from the reprogrammed foundation model. TDMR is compatible with different pre-trained model types (CNN, transformer or their mix) and limited target data, and promotes the wide applications of vision foundation models to downstream tasks in a cost-effective manner. Extensive experiments on different downstream classification tasks and target model structures demonstrate the effectiveness of our methods with both CNNs and transformer foundation models.

arXiv:2304.02168공식 코드 · Papers with Code

I2I: Initializing Adapters with Improvised Knowledge

Adapters present a promising solution to the catastrophic forgetting problem in continual learning. However, training independent Adapter modules for every new task misses an opportunity for cross-task knowledge transfer. We propose Improvise to Initialize (I2I), a continual learning algorithm that initializes Adapters for incoming tasks by distilling knowledge from previously-learned tasks' Adapters. We evaluate I2I on CLiMB, a multimodal continual learning benchmark, by conducting experiments on sequences of visual question answering tasks. Adapters trained with I2I consistently achieve better task accuracy than independently-trained Adapters, demonstrating that our algorithm facilitates knowledge transfer between task Adapters. I2I also results in better cross-task knowledge transfer than the state-of-the-art AdapterFusion without incurring the associated parametric cost.

arXiv:2304.01354공식 코드 · Papers with Code

Functional Knowledge Transfer with Self-supervised Representation Learning

This work investigates the unexplored usability of self-supervised representation learning in the direction of functional knowledge transfer. In this work, functional knowledge transfer is achieved by joint optimization of self-supervised learning pseudo task and supervised learning task, improving supervised learning task performance. Recent progress in self-supervised learning uses a large volume of data, which becomes a constraint for its applications on small-scale datasets. This work shares a simple yet effective joint training framework that reinforces human-supervised task learning by learning self-supervised representations just-in-time and vice versa. Experiments on three public datasets from different visual domains, Intel Image, CIFAR, and APTOS, reveal a consistent track of performance improvements on classification tasks during joint optimization. Qualitative analysis also supports the robustness of learnt representations. Source code and trained models are available on GitHub.

arXiv:2304.01731공식 코드 · Papers with Code

Selective Knowledge Sharing for Privacy-Preserving Federated Distillation without A Good Teacher

While federated learning is promising for privacy-preserving collaborative learning without revealing local data, it remains vulnerable to white-box attacks and struggles to adapt to heterogeneous clients. Federated distillation (FD), built upon knowledge distillation--an effective technique for transferring knowledge from a teacher model to student models--emerges as an alternative paradigm, which provides enhanced privacy guarantees and addresses model heterogeneity. Nevertheless, challenges arise due to variations in local data distributions and the absence of a well-trained teacher model, which leads to misleading and ambiguous knowledge sharing that significantly degrades model performance. To address these issues, this paper proposes a selective knowledge sharing mechanism for FD, termed Selective-FD. It includes client-side selectors and a server-side selector to accurately and precisely identify knowledge from local and ensemble predictions, respectively. Empirical studies, backed by theoretical insights, demonstrate that our approach enhances the generalization capabilities of the FD framework and consistently outperforms baseline methods.

arXiv:2304.01404

Adaptive Defective Area Identification in Material Surface Using Active Transfer Learning-based Level Set Estimation

In material characterization, identifying defective areas on a material surface is fundamental. The conventional approach involves measuring the relevant physical properties point-by-point at the predetermined mesh grid points on the surface and determining the area at which the property does not reach the desired level. To identify defective areas more efficiently, we propose adaptive mapping methods in which measurement resources are used preferentially to detect the boundaries of defective areas. We interpret this problem as an active-learning (AL) of the level set estimation (LSE) problem. The goal of AL-based LSE is to determine the level set of the physical property function defined on the surface with as small number of measurements as possible. Furthermore, to handle the situations in which materials with similar specifications are repeatedly produced, we introduce a transfer learning approach so that the information of previously produced materials can be effectively utilized. As a proof-of-concept, we applied the proposed methods to the red-zone estimation problem of silicon wafers and demonstrated that we could identify the defective areas with significantly lower measurement costs than those of conventional methods.

arXiv:2304.01524

FisHook -- An Optimized Approach to Marine Specie Classification using MobileNetV2

Marine ecosystems are vital for the planet's health, but human activities such as climate change, pollution, and overfishing pose a constant threat to marine species. Accurate classification and monitoring of these species can aid in understanding their distribution, population dynamics, and the impact of human activities on them. However, classifying marine species can be challenging due to their vast diversity and the complex underwater environment. With advancements in computer performance and GPU-based computing, deep-learning algorithms can now efficiently classify marine species, making it easier to monitor and manage marine ecosystems. In this paper, we propose an optimization to the MobileNetV2 model to achieve a 99.83% average validation accuracy by highlighting specific guidelines for creating a dataset and augmenting marine species images. This transfer learning algorithm can be deployed successfully on a mobile application for on-site classification at fisheries.

arXiv:2304.01295공식 코드 · Papers with Code

Efficiently Aligned Cross-Lingual Transfer Learning for Conversational Tasks using Prompt-Tuning

Cross-lingual transfer of language models trained on high-resource languages like English has been widely studied for many NLP tasks, but focus on conversational tasks has been rather limited. This is partly due to the high cost of obtaining non-English conversational data, which results in limited coverage. In this work, we introduce XSGD for cross-lingual alignment pretraining, a parallel and large-scale multilingual conversation dataset that we created by translating the English-only Schema-Guided Dialogue (SGD) dataset (Rastogi et al., 2020) into 105 other languages. XSGD contains approximately 330k utterances per language. To facilitate aligned cross-lingual representations, we develop an efficient prompt-tuning-based method for learning alignment prompts. We also investigate two different classifiers: NLI-based and vanilla classifiers, and test cross-lingual capability enabled by the aligned prompts. We evaluate our model's cross-lingual generalization capabilities on two conversation tasks: slot-filling and intent classification. Our results demonstrate the strong and efficient modeling ability of NLI-based classifiers and the large cross-lingual transfer improvements achieved by our aligned prompts, particularly in few-shot settings. In addition, we highlight the nice results of our approach compared to LLMs such as text-davinci-003 and ChatGPT in both zero-shot and few-shot settings. While LLMs exhibit impressive performance in English, their cross-lingual capabilities in other languages, particularly low-resource languages, are limited.

arXiv:2304.01279공식 코드 · Papers with Code

Long-Tailed Visual Recognition via Self-Heterogeneous Integration with Knowledge Excavation

Deep neural networks have made huge progress in the last few decades. However, as the real-world data often exhibits a long-tailed distribution, vanilla deep models tend to be heavily biased toward the majority classes. To address this problem, state-of-the-art methods usually adopt a mixture of experts (MoE) to focus on different parts of the long-tailed distribution. Experts in these methods are with the same model depth, which neglects the fact that different classes may have different preferences to be fit by models with different depths. To this end, we propose a novel MoE-based method called Self-Heterogeneous Integration with Knowledge Excavation (SHIKE). We first propose Depth-wise Knowledge Fusion (DKF) to fuse features between different shallow parts and the deep part in one network for each expert, which makes experts more diverse in terms of representation. Based on DKF, we further propose Dynamic Knowledge Transfer (DKT) to reduce the influence of the hardest negative class that has a non-negligible impact on the tail classes in our MoE framework. As a result, the classification accuracy of long-tailed data can be significantly improved, especially for the tail classes. SHIKE achieves the state-of-the-art performance of 56.3%, 60.3%, 75.4%, and 41.9% on CIFAR100-LT (IF100), ImageNet-LT, iNaturalist 2018, and Places-LT, respectively.

arXiv:2304.01093

Demonstration of a Standalone, Descriptive, and Predictive Digital Twin of a Floating Offshore Wind Turbine

Digital Twins bring several benefits for planning, operation, and maintenance of remote offshore assets. In this work, we explain the digital twin concept and the capability level scale in the context of wind energy. Furthermore, we demonstrate a standalone digital twin, a descriptive digital twin, and a prescriptive digital twin of an operational floating offshore wind turbine. The standalone digital twin consists of the virtual representation of the wind turbine and its operating environment. While at this level the digital twin does not evolve with the physical turbine, it can be used during the planning-, design-, and construction phases. At the next level, the descriptive digital twin is built upon the standalone digital twin by enhancing the latter with real data from the turbine. All the data is visualized in virtual reality for informed decision-making. Besides being used for data bundling and visualization, the descriptive digital twin forms the basis for diagnostic, predictive, prescriptive, and autonomous tools. A predictive digital twin is created through the use of weather forecasts, neural networks, and transfer learning. Finally, digital twin technology is discussed in a much wider context of ocean engineering.

arXiv:2304.00070

HybridCVLNet: A Hybrid CSI Feedback System and its Domain Adaptation

Deep Learning (DL)-based channel state information (CSI) feedback is a promising technique for the transmitter to accurately acquire the CSI of massive multiple-input multiple-output (MIMO) systems. As critical concerns about DL-based physical layer applications, the intra-domain generalizability affected by dataset bias and inter-domain robustness in data drift remain challenging. Therefore, we build on a Hybrid Complex-Valued Lightweight framework, namely the HybridCVLNet, capable of overcoming the dataset bias with regularized hybrid structure and codeword. Meanwhile, a corresponding transductive-based hybrid domain adaptation scheme is proposed to tackle the inter-domain data drift. The experiment verifies that HybridCVLNet achieves stable generalizability and performance gain over the state-of-the-art (SOTA) feedback schemes in an intra-domain heterogeneous dataset. In addition, its transductive-based hybrid domain adaptation scheme is more efficient and superior to the inductive-based transfer learning methods under two inter-domain online re-optimization settings.