Fine-tuning Approach Research Articles

Enhancing domain-specific text generation for power grid maintenance with P2FT.

The digitization of operation and maintenance in the intelligent power grid equipment relies on a diverse array of information for smart decision-making. In the domain of intelligent decision generation, proficiency is contingent upon extensive learning from copious amounts of text. This necessitates not only robust processing capabilities but also a high level of specialization. In addressing situations where authorization is lacking, pre-trained language models (PLMs) have already provided ideas when confronted with specialized domains or tasks. In consideration of the complexity of textual content in the field of the power grid, which encompasses a multitude of specialized knowledge and involves an abundance of proprietary terminology, we have undertaken an exploration of pre-trained model specialization using the power grid domain as an example, specifically for the task of generating maintenance strategies. A two-stage fine-tuning approach (P2FT) is employed, utilizing a large-scale pre-training model specifically designed for natural language processing. The efficacy and practical value of this method were evaluated through multiple metrics, juxtaposed with other advanced approaches involving low-parameter or parameter-free fine-tuning methods. Through a meticulous analysis and validation of experimental outcomes, we have corroborated the feasibility and practical application value of employing this approach for pre-trained model specialization. Additionally, it has furnished valuable guidance for text generation within both the Chinese language domain and the power grid domain.

Knowledge Transfer from High-Resource to Low-Resource Programming Languages for Code LLMs

Over the past few years, Large Language Models of Code (Code LLMs) have started to have a significant impact on programming practice. Code LLMs are also emerging as building blocks for research in programming languages and software engineering. However, the quality of code produced by a Code LLM varies significantly by programming language. Code LLMs produce impressive results on high-resource programming languages that are well represented in their training data (e.g., Java, Python, or JavaScript), but struggle with low-resource languages that have limited training data available (e.g., OCaml, Racket, and several others). This paper presents an effective approach for boosting the performance of Code LLMs on low-resource languages using semi-synthetic data. Our approach, called MultiPL-T, generates high-quality datasets for low-resource languages, which can then be used to fine-tune any pretrained Code LLM. MultiPL-T translates training data from high-resource languages into training data for low-resource languages in the following way. 1) We use a Code LLM to synthesize unit tests for commented code from a high-resource source language, filtering out faulty tests and code with low test coverage. 2) We use a Code LLM to translate the code from the high-resource source language to a target low-resource language. This gives us a corpus of candidate training data in the target language, but many of these translations are wrong. 3) We use a lightweight compiler to compile the test cases generated in (1) from the source language to the target language, which allows us to filter our obviously wrong translations. The result is a training corpus in the target low-resource language where all items have been validated with test cases. We apply this approach to generate tens of thousands of new, validated training items for five low-resource languages: Julia, Lua, OCaml, R, and Racket, using Python as the source high-resource language. Furthermore, we use an open Code LLM (StarCoderBase) with open training data (The Stack), which allows us to decontaminate benchmarks, train models without violating licenses, and run experiments that could not otherwise be done. Using datasets generated with MultiPL-T, we present fine-tuned versions of StarCoderBase and Code Llama for Julia, Lua, OCaml, R, and Racket that outperform other fine-tunes of these base models on the natural language to code task. We also present Racket fine-tunes for two very recent models, DeepSeek Coder and StarCoder2, to show that MultiPL-T continues to outperform other fine-tuning approaches for low-resource languages. The MultiPL-T approach is easy to apply to new languages, and is significantly more efficient and effective than alternatives such as training longer.

Identification of perceived sentences using deep neural networks in EEG

Objetive. Decoding speech from brain activity can enable communication for individuals with speech disorders. Deep neural networks (DNNs) have shown great potential for speech decoding applications. However, the limited availability of large datasets containing neural recordings from speech-impaired subjects poses a challenge. Leveraging data from healthy participants can mitigate this limitation and expedite the development of speech neuroprostheses while minimizing the need for patient-specific training data.Approach. In this study, we collected a substantial dataset consisting of recordings from 56 healthy participants using 64 EEG channels. Multiple neural networks were trained to classify perceived sentences in the Spanish language using subject-independent, mixed-subjects, and fine-tuning approaches. The dataset has been made publicly available to foster further research in this area.Main results. Our results demonstrate a remarkable level of accuracy in distinguishing sentence identity across 30 classes, showcasing the feasibility of training DNNs to decode sentence identity from perceived speech using EEG. Notably, the subject-independent approach rendered accuracy comparable to the mixed-subjects approach, although with higher variability among subjects. Additionally, our fine-tuning approach yielded even higher accuracy, indicating an improved capability to adapt to individual subject characteristics, which enhances performance. This suggests that DNNs have effectively learned to decode universal features of brain activity across individuals while also being adaptable to specific participant data. Furthermore, our analyses indicate that EEGNet and DeepConvNet exhibit comparable performance, outperforming ShallowConvNet for sentence identity decoding. Finally, our Grad-CAM visualization analysis identifies key areas influencing the network's predictions, offering valuable insights into the neural processes underlying language perception and comprehension.Significance. These findings advance our understanding of EEG-based speech perception decoding and hold promise for the development of speech neuroprostheses, particularly in scenarios where subjects cannot provide their own training data.

Fine-grained detoxification framework via instance-level prefixes for large language models

Large Language Models (LLMs) have demonstrated remarkable performance across various natural language processing (NLP) tasks. However, their practical usability is often compromised by a propensity to generate toxic content, such as insults, threats, and profanity, particularly in response to adversarial prompts. Several fine-tuning and decoding approaches have been employed to address this challenge to mitigate toxicity. Nevertheless, these methods typically necessitate additional resources, such as high-quality training data or auxiliary models, thereby incurring extra costs. In this paper, we propose a novel detoxification framework, Fine-Grained Detoxification via Instance-Level Prefixes (FGDILP), which effectively mitigates the generation of toxic text without incurring additional training costs. Specifically, FGDILP leverages contextualized representations in attention space by contrasting a positive prefix-prepended prompt against multiple negative prefix-prepended prompts at the instance level. This methodology facilitates the construction of fine-grained subtoxicity vectors, which are subsequently fused to adjust the original generation pathway when responding to raw prompts. Our results demonstrate that FGDILP enables controlled text generation concerning detoxification at both the utterance and context levels. While our method slightly impacts generation fluency and diversity, it significantly outperforms prompt-based baselines regarding detoxification effectiveness. Our code is available at https://github.com/xinykou/FGDILP.

Deep Learning-Based Fine-Tuning Approach of Coarse Registration for Ear-Nose-Throat (ENT) Surgical Navigation Systems.

Accurate registration between medical images and patient anatomy is crucial for surgical navigation systems in minimally invasive surgeries. This study introduces a novel deep learning-based refinement step to enhance the accuracy of surface registration without disrupting established workflows. The proposed method integrates a machine learning model between conventional coarse registration and ICP fine registration. A deep-learning model was trained using simulated anatomical landmarks with introduced localization errors. The model architecture features global feature-based learning, an iterative prediction structure, and independent processing of rotational and translational components. Validation with silicon-masked head phantoms and CT imaging compared the proposed method to both conventional registration and a recent deep-learning approach. The results demonstrated significant improvements in target registration error (TRE) across different facial regions and depths. The average TRE for the proposed method (1.58 ± 0.52 mm) was significantly lower than that of the conventional (2.37 ± 1.14 mm) and previous deep-learning (2.29 ± 0.95 mm) approaches (p < 0.01). The method showed a consistent performance across various facial regions and enhanced registration accuracy for deeper areas. This advancement could significantly enhance precision and safety in minimally invasive surgical procedures.

A Novel Approach for Stratifying Pulmonary Edema Severity on Chest X-ray via Dual-Mechanic Self-Learning and Bidirectional Multi-Modal Cross-Attention Algorithms

Abstract Accurate assessment of pulmonary edema severity in acute decompensated congestive heart failure (CHF) patients is vital for treatment decisions. Traditional methods face challenges due to the complexity of chest X-ray (CXR) and unstructured radiology reports. We proposed a method combining self-supervised learning and multimodal cross-attention to address these challenges. Dual-mechanic self-supervised pre-training enhances feature extraction using contrastive learning between text and image features, and generative learning between images. A bidirectional multi-modal cross-attention model integrates image and text information for fine-tuning, improving model performance. Four CXR datasets consisting of 519, 437 images were used for pre-training; 1200 randomly selected image-text pairs were used for fine-tuning and partitioned into train, validation, and test sets at 3: 1: 1. Ablation studies for pre-training and fine-tuning approaches indicated their practicality as evidenced by the optimal macro F1 score of 0.667 and optimal macro-AUC of 0.904. It also outperformed other state-of-the-art multi-modality methods. The novel approach could accurately assess pulmonary edema severity, offering crucial support for CHF patient management.

Comparison of Hybrid Machine Learning Approaches for Surrogate Modeling Part Shrinkage in Injection Molding.

Machine learning (ML) methods present a valuable opportunity for modeling the non-linear behavior of the injection molding process. They have the potential to predict how various process and material parameters affect the quality of the resulting parts. However, the dynamic nature of the injection molding process and the challenges associated with collecting process data remain significant obstacles for the application of ML methods. To address this, within this study, hybrid approaches are compared that combine process data with additional process knowledge, such as constitutive equations and high-fidelity numerical simulations. The hybrid modeling approaches include feature learning, fine-tuning, delta-modeling, preprocessing, and using physical constraints, as well as combinations of the individual approaches. To train and validate the hybrid models, both the experimental and simulated shrinkage data of an injection-molded part are utilized. While all hybrid approaches outperform the purely data-based model, the fine-tuning approach yields the best result in the simulation setting. The combination of calibrating a physical model (feature learning) and incorporating it implicitly into the training process (physical constraints) outperforms the other approaches in the experimental setting.

Optimizing Food101 Classification with Transfer Learning: A Fine-Tuning Approach Using EfficientNetB0

Much research has been done on the classification of the food101 dataset, but much of this research which achieved an accuracy score of more than 90% explores heavyweight architecture such as EfficientNetB7, Visual Geometry Group19, ResNet-200, Inception v4, DenseNet-201, ResNeXt-101, MobileNet v3 and many more. This study explores the classification of the Food101 dataset using the EfficientNetB0 architecture, a lightweight architecture. Compared to other popular CNN architecture, EfficientNetB0 has relatively small parameters, which makes it computationally efficient and suitable for deployment on resource-constraint environments. The research aims to balance model accuracy and computational efficiency, addressing the need for resource-constrained environments. Five experiments were conducted while varying the number of fine-tuned layers. Results demonstrate that the fine-tuned EfficientNetB0 model achieves an accuracy score of accuracy score of 97.54%, Top_k_categorical accuracy of 99.89%, precision of 98.21%, and recall of 97.02% in just 5 epochs. This research will significantly contribute to the field of transfer learning by developing specialized models that excel in target tasks. Besides, it will advance dietary monitoring, food logging, and health-related technologies, enabling more accessible and practical solutions for consumers. However, the optimal number of layers to fine-tune for achieving perfect accuracy with EfficientNetB0 remains uncertain. It often involves trial and error to determine the best configuration for optimal results, presenting an opportunity for future research.

Exploring the Trade-Off between generalist and specialized Models: A center-based comparative analysis for glioblastoma segmentation

IntroductionInherent variations between inter-center data can undermine the robustness of segmentation models when applied at a specific center (dataset shift). We investigated whether specialized center-specific models are more effective compared to generalist models based on multi-center data, and how center-specific data could enhance the performance of generalist models within a particular center using a fine-tuning transfer learning approach. For this purpose, we studied the dataset shift at center level and conducted a comparative analysis to assess the impact of data source on glioblastoma segmentation models. Methods & MaterialsThe three key components of dataset shift were studied: prior probability shift—variations in tumor size or tissue distribution among centers; covariate shift—inter-center MRI alterations; and concept shift—different criteria for tumor segmentation. BraTS 2021 dataset was used, which includes 1251 cases from 23 centers. Thereafter, 155 deep-learning models were developed and compared, including 1) generalist models trained with multi-center data, 2) specialized models using only center-specific data, and 3) fine-tuned generalist models using center-specific data. ResultsThe three key components of dataset shift were characterized. The amount of covariate shift was substantial, indicating large variations in MR imaging between different centers. Glioblastoma segmentation models tend to perform best when using data from the application center. Generalist models, trained with over 700 samples, achieved a median Dice score of 88.98%. Specialized models surpassed this with 200 cases, while fine-tuned models outperformed with 50 cases. ConclusionsThe influence of dataset shift on model performance is evident. Fine-tuned and specialized models, utilizing data from the evaluated center, outperform generalist models, which rely on data from other centers. These approaches could encourage medical centers to develop customized models for their local use, enhancing the accuracy and reliability of glioblastoma segmentation in a context where dataset shift is inevitable.

A New Adapter Tuning of Large Language Model for Chinese Medical Named Entity Recognition

ABSTRACT Named entity recognition (NER) is a crucial step in extracting medical information from Chinese text, and fine-tuning large language models (LLMs) for this task is an effective approach. However, full parameter fine-tuning can potentially damage the model’s original parameters, resulting in catastrophic forgetting. To overcome this challenge, we introduce a novel adapter-based fine-tuning approach. Our adapter is integrated into the first and last transformers of the LLM, operating in parallel to the feed-forward network (FFN), following multi-head attention. It mirrors the FFN’s structure and uses the FFN’s weights for initializing. Additionally, to further enhance performance, we incorporate prefix embeddings into the first and last transformers. Our experiments on the Chinese medical NER benchmark demonstrate that our adapter, combined with prefix embeddings, achieves the highest F1-score of 65.90%, surpassing prompt templates (21.99%), in-context learning (18.65%), P-tuning (63.03%), and the benchmark for the Chinese medical NER task (62.40%). These results indicate that our adapter effectively fine-tunes the LLM for Chinese medical NER while preserving the original parameters.

Application of deep learning methods for beam size control during user operation at the Advanced Light Source

Past research at the Advanced Light Source (ALS) provided a proof-of-principle demonstration that deep learning methods could be effectively employed to compensate for the significant perturbations to the transverse electron beam size induced by user-controlled adjustments of the insertion devices. However, incorporating these methods into the ALS’ daily operations has faced notable challenges. The complexity of the system’s operational requirements and the significant upkeep demands has restricted their sustained application during user operation. Here, we introduce the development of a more robust neural network (NN)-based algorithm that utilizes a novel online fine-tuning approach and its systematic integration into the day-to-day machine operations. Our analysis emphasizes the process of NN model selection, demonstrates the superior performance of the NN-based method over traditional feedback methods, and examines the effectiveness and resilience of the new algorithm during user-operation scenarios. Published by the American Physical Society 2024

Three-Stage Framework for Accurate Pediatric Chest X-ray Diagnosis Using Self-Supervision and Transfer Learning on Small Datasets.

Pediatric respiratory disease diagnosis and subsequent treatment require accurate and interpretable analysis. A chest X-ray is the most cost-effective and rapid method for identifying and monitoring various thoracic diseases in children. Recent developments in self-supervised and transfer learning have shown their potential in medical imaging, including chest X-ray areas. In this article, we propose a three-stage framework with knowledge transfer from adult chest X-rays to aid the diagnosis and interpretation of pediatric thorax diseases. We conducted comprehensive experiments with different pre-training and fine-tuning strategies to develop transformer or convolutional neural network models and then evaluate them qualitatively and quantitatively. The ViT-Base/16 model, fine-tuned with the CheXpert dataset, a large chest X-ray dataset, emerged as the most effective, achieving a mean AUC of 0.761 (95% CI: 0.759-0.763) across six disease categories and demonstrating a high sensitivity (average 0.639) and specificity (average 0.683), which are indicative of its strong discriminative ability. The baseline models, ViT-Small/16 and ViT-Base/16, when directly trained on the Pediatric CXR dataset, only achieved mean AUC scores of 0.646 (95% CI: 0.641-0.651) and 0.654 (95% CI: 0.648-0.660), respectively. Qualitatively, our model excels in localizing diseased regions, outperforming models pre-trained on ImageNet and other fine-tuning approaches, thus providing superior explanations. The source code is available online and the data can be obtained from PhysioNet.

Enhancing molecular property prediction with auxiliary learning and task-specific adaptation

Pretrained Graph Neural Networks have been widely adopted for various molecular property prediction tasks. Despite their ability to encode structural and relational features of molecules, traditional fine-tuning of such pretrained GNNs on the target task can lead to poor generalization. To address this, we explore the adaptation of pretrained GNNs to the target task by jointly training them with multiple auxiliary tasks. This could enable the GNNs to learn both general and task-specific features, which may benefit the target task. However, a major challenge is to determine the relatedness of auxiliary tasks with the target task. To address this, we investigate multiple strategies to measure the relevance of auxiliary tasks and integrate such tasks by adaptively combining task gradients or by learning task weights via bi-level optimization. Additionally, we propose a novel gradient surgery-based approach, Rotation of Conflicting Gradients (RCGrad\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\mathop {\ exttt{RCGrad}}\\limits$$\\end{document}), that learns to align conflicting auxiliary task gradients through rotation. Our experiments with state-of-the-art pretrained GNNs demonstrate the efficacy of our proposed methods, with improvements of up to 7.7% over fine-tuning. This suggests that incorporating auxiliary tasks along with target task fine-tuning can be an effective way to improve the generalizability of pretrained GNNs for molecular property prediction.Scientific contributionWe introduce a novel framework for adapting pretrained GNNs to molecular tasks using auxiliary learning to address the critical issue of negative transfer. Leveraging novel gradient surgery techniques such as RCGrad\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\mathop {\ exttt{RCGrad}}\\limits$$\\end{document}, the proposed adaptation framework represents a significant departure from the dominant pretraining fine-tuning approach for molecular GNNs. Our contributions are significant for drug discovery research, especially for tasks with limited data, filling a notable gap in the efficient adaptation of pretrained models for molecular GNNs.

Deep neural networks for external corrosion classification in industrial above-ground storage tanks

This paper uses transfer learning and fine-tuning approaches to explore EfficientNet smaller models for a multiclassification task of corrosion types in above-ground storage tanks. Data augmentation was used to increment the data an oil and gas company provided, reaching a dataset of around 5000 images. The images were stored in Google Drive and imported into Colab to train the models using TensorFlow and Keras. After the hyperparameters’ tuning a transfer learning model was selected and explored with fine-tuning. The EfficientNetB0 model delivered from fine-tuning accomplished 94 % performance. This work is the first attempt to deploy an artificial vision automatic tool for being implemented during tank inspection in the industrial sector. In further development, this model can be coupled with one based on object detection to remotely identify failures due to external corrosion issues, improving safety and reliability in oil and gas operations.

Artificial intelligence co-piloted auditing

This paper proposes the concept of artificial intelligence co-piloted auditing, emphasizing the collaborative potential of auditors and foundation models in the auditing domain. The paper discusses the future relationship and interactions of human auditors and AI, imagining an audit setup where auditors’ capabilities are enhanced through artificial intelligence across a variety of audit tasks. To exemplify the potential of this co-piloted audit paradigm, we illustrate a systematic fine-tuning approach to foundation models using Chain-of-Thought prompting. This study showcases how foundation models can work as collaborators flexibly with auditors, enabling the model to accurately identify transactions from instructions. This study provides a detailed description of the formulated prompt protocols and the corresponding responses generated by ChatGPT, ensuring reproducibility. We envision this work as an initial step towards the widespread implementation of co-piloted auditing, paving the way for more efficient, accurate, and insightful audit procedures.

Contextualized dynamic meta embeddings based on Gated CNNs and self-attention for Arabic machine translation

PurposeThe paper aims to enhance Arabic machine translation (MT) by proposing novel approaches: (1) a dimensionality reduction technique for word embeddings tailored for Arabic text, optimizing efficiency while retaining semantic information; (2) a comprehensive comparison of meta-embedding techniques to improve translation quality; and (3) a method leveraging self-attention and Gated CNNs to capture token dependencies, including temporal and hierarchical features within sentences, and interactions between different embedding types. These approaches collectively aim to enhance translation quality by combining different embedding schemes and leveraging advanced modeling techniques.Design/methodology/approachRecent works on MT in general and Arabic MT in particular often pick one type of word embedding model. In this paper, we present a novel approach to enhance Arabic MT by addressing three key aspects. Firstly, we propose a new dimensionality reduction technique for word embeddings, specifically tailored for Arabic text. This technique optimizes the efficiency of embeddings while retaining their semantic information. Secondly, we conduct an extensive comparison of different meta-embedding techniques, exploring the combination of static and contextual embeddings. Through this analysis, we identify the most effective approach to improve translation quality. Lastly, we introduce a novel method that leverages self-attention and Gated convolutional neural networks (CNNs) to capture token dependencies, including temporal and hierarchical features within sentences, as well as interactions between different types of embeddings. Our experimental results demonstrate the effectiveness of our proposed approach in significantly enhancing Arabic MT performance. It outperforms baseline models with a BLEU score increase of 2 points and achieves superior results compared to state-of-the-art approaches, with an average improvement of 4.6 points across all evaluation metrics.FindingsThe proposed approaches significantly enhance Arabic MT performance. The dimensionality reduction technique improves the efficiency of word embeddings while preserving semantic information. Comprehensive comparison identifies effective meta-embedding techniques, with the contextualized dynamic meta-embeddings (CDME) model showcasing competitive results. Integration of Gated CNNs with the transformer model surpasses baseline performance, leveraging both architectures' strengths. Overall, these findings demonstrate substantial improvements in translation quality, with a BLEU score increase of 2 points and an average improvement of 4.6 points across all evaluation metrics, outperforming state-of-the-art approaches.Originality/valueThe paper’s originality lies in its departure from simply fine-tuning the transformer model for a specific task. Instead, it introduces modifications to the internal architecture of the transformer, integrating Gated CNNs to enhance translation performance. This departure from traditional fine-tuning approaches demonstrates a novel perspective on model enhancement, offering unique insights into improving translation quality without solely relying on pre-existing architectures. The originality in dimensionality reduction lies in the tailored approach for Arabic text. While dimensionality reduction techniques are not new, the paper introduces a specific method optimized for Arabic word embeddings. By employing independent component analysis (ICA) and a post-processing method, the paper effectively reduces the dimensionality of word embeddings while preserving semantic information which has not been investigated before especially for MT task.

Data-efficient comfort modeling: Active transfer learning for predicting personal thermal comfort using limited data

Personal comfort models are used to predict thermal comfort responses at the individual level rather than predicting the average thermal comfort responses for large populations. These models, data-driven in nature, need to be trained on large amounts of occupant comfort feedback and sensor data to achieve accurate predictions. However, collecting such data is often expensive and labor-intensive in reality. To address this, we proposed a data-efficient active transfer learning (ATL) framework to improve the performance of personal comfort models under limited data. To demonstrate the validity of this framework, we developed a base Convolutional Neural Network-Long Short-term Memory (CNNLSTM) model alongside two transfer learning models utilizing feature extraction (TL-CNNLSTM-FE) and fine-tuning (TL-CNNLSTM-FE) approaches, enhanced by a novel active learning strategy. Using these models, three comfort prediction tasks (i.e., thermal preference, thermal acceptability, and air movement preference) were performed by transferring the knowledge from the ASHRAE Global Thermal Comfort Database II to a limited dataset collected in the tropics. Empirical results indicate that the active transfer learning framework proposed was able to consistently outperform the base and transfer learning models using only less than 10% of the training data for all personal comfort tasks, highlighting the effectiveness of this strategy. The implications of this work are especially useful for the research community working on the practical applications of data-efficient machine learning approaches for personal thermal comfort predictions.

Improving unbalanced image classification through fine-tuning method of reinforcement learning

Image classification, especially unbalanced image classification, holds considerable promise for practical applications. Existing research focuses mainly on enhancing the effectiveness of classifiers through approaches such as data resampling and loss function adjustments. To date, most of the approaches address the class unbalanced issue by transforming unbalanced data distributions into balanced [id=SecondEdit]ones. Consequently, a critical challenge is to directly develop high-performance classifiers that are adaptive to diverse unbalanced data distributions. In this paper, for unbalanced image classification tasks, we propose a Reinforcement Learning Fine-tuning approach to Unbalanced image Classification (RLF-UC). Specifically, we train classification pretraining models on [d=FirstEdit]fivethree unbalanced datasetsid=SecondEdit], and train [d=FirstEdit]the correspondinga reward function model designed to optimize fine-tuning policy. Then, we train a reinforcement learning fine-tuning classification model and optimize its policy to maximize the cumulative expected reward. Finally, we guide the model to prioritize minority category knowledge and incorporate distribution distance constraints, which are derived from the disparities between the [d=SecondEdit]pre-trainedpretrained model and the fine-tuning model to dynamically adjust the fine-tuning classifier model. Experimental results on [d=FirstEdit]fivethree reprocessed unbalanced image datasets demonstrate that our RLF-UC method provides comparable or better classification and generalization capabilities than other baselines.

Practical Techniques for Vision-Language Segmentation Model in Remote Sensing

Abstract. Traditional semantic segmentation models often struggle with poor generalizability in zero-shot scenarios such as recognizing attributes unseen in the training labels. On the other hands, language-vision models (VLMs) have shown promise in improving performance on zero-shot tasks by leveraging semantic information from textual inputs and fusing this information with visual features. However, existing VLM-based methods do not perform as effectively on remote sensing data due to the lack of such data in their training datasets. In this paper, we introduce a two-stage fine-tuning approach for a VLM-based segmentation model using a large remote sensing image-caption dataset, which we created using an existing image-caption model. Additionally, we propose a modified decoder and a visual prompt technique using a saliency map to enhance segmentation results. Through these methods, we achieve superior segmentation performance on remote sensing data, demonstrating the effectiveness of our approach.

Iterative learning for efficient additive mass production

Material extrusion could enable on-demand production of complex and personalized parts but is limited by low reliability, particularly in higher-volume production. Machine learning-based methods may enhance reliability, but are often themselves insufficiently reliable for use in production. Foundation artificial intelligence models have enabled significant improvements in performance across many tasks. Here, a vision-based control system is reported, coupling active learning and uncertainty awareness with a foundation model to continually learn to build a specific part better. The resulting framework is called Iterative Learning, as it improves performance by learning from its own errors during repeated build cycles of the same part. The iterative learning approach is shown to enable robust error detection and correction while being more space, time and computationally efficient compared to a naive fine-tuning approach. This provides a path showing how foundation models may be adapted to enhance reliability across a wider range of additive manufacturing processes.