Machine learning books and papers

با عرض سلام مقاله ی زیر تماما نگارش شده و اماده سابمیت از دوستان کسی خواست نفرات ۳ و ۴ اش خالی هست. IEEE Geoscience and Remote Sensing Letter Impact factor 4 CiteScore 7.6 ------------------------------ Title: Enhanced-HisSegNet: An Enhanced Histagram Layered Segmentation Network for SAR Image-based Flood Segmentation ------------------------------ Abstract: Floods are among the most frequent natural disasters, causing loss of life and significant economic and environmental damage, with direct impacts on agriculture, urban infrastructure, and transportation networks. Therefore, it is crucial to accurately and efficiently identify flooded areas in the aftermath of such events. Synthetic Aperture Radar (SAR) imagery plays a vital role in this process, as water surfaces reflect less microwave energy compared to land due to their smooth texture and low surface roughness. In this study, we present a multimodal fusion strategy that enhances the existing model by Turkmenli et al. [1] through the integration of fine-tuned histograms and Deep Neural Networks (DNNs) for improved flood mapping. Specifically, we introduce fine-tuned histogram extraction layers designed for SAR data, which are integrated into Deep Segmentation Neural Networks (DSNNs). The model was tested on two real SAR datasets, with cross-dataset validation using an external cohort, representing a second innovation in our approach. Experimental results demonstrate that our model, with fine-tuned histogram layers, outperforms previous approaches by up to 4% in intersection over union (IoU) and provides a comprehensive evaluation through metrics such as Precision, Recall, Average Precision (AP), Mean Average Precision (mAP), False Positive Rate (FPR), and Mean Average Recall (mAR). Importantly, these improvements come with minimal additional learnable parameters. The code for this work will be made available at https://github.com/Mohsena1990/Enhanced-HistSegNet @Raminmousa @Machine_learn @Paper4monry

با عرض سلام مقاله ی زیر تماما نگارش شده و اماده سابمیت از دوستان کسی خواست نفرات ۲ تا ۴ اش خالی هست. IEEE Geoscience and Remote Sensing Letter Impact factor 4 CiteScore 7.6 ------------------------------ Title: Enhanced-HisSegNet: An Enhanced Histagram Layered Segmentation Network for SAR Image-based Flood Segmentation ------------------------------ Abstract: Floods are among the most frequent natural disasters, causing loss of life and significant economic and environmental damage, with direct impacts on agriculture, urban infrastructure, and transportation networks. Therefore, it is crucial to accurately and efficiently identify flooded areas in the aftermath of such events. Synthetic Aperture Radar (SAR) imagery plays a vital role in this process, as water surfaces reflect less microwave energy compared to land due to their smooth texture and low surface roughness. In this study, we present a multimodal fusion strategy that enhances the existing model by Turkmenli et al. [1] through the integration of fine-tuned histograms and Deep Neural Networks (DNNs) for improved flood mapping. Specifically, we introduce fine-tuned histogram extraction layers designed for SAR data, which are integrated into Deep Segmentation Neural Networks (DSNNs). The model was tested on two real SAR datasets, with cross-dataset validation using an external cohort, representing a second innovation in our approach. Experimental results demonstrate that our model, with fine-tuned histogram layers, outperforms previous approaches by up to 4% in intersection over union (IoU) and provides a comprehensive evaluation through metrics such as Precision, Recall, Average Precision (AP), Mean Average Precision (mAP), False Positive Rate (FPR), and Mean Average Recall (mAR). Importantly, these improvements come with minimal additional learnable parameters. The code for this work will be made available at https://github.com/Mohsena1990/Enhanced-HistSegNet @Raminmousa @Machine_learn @Paper4monry

با عرض سلام دوستان اين گروه ها كامل پرشده بجز بخش طبقه بندي پزشكي كه نفر پنجم يه گروه جا هست . @Raminmousa

LLM based Multi-Agent methods 🖥 Github: https://github.com/AgnostiqHQ/multi-agent-llm 📕 Paper: https://arxiv.org/abs/2409.12618v1 🤗 Dataset: https://paperswithcode.com/dataset/hotpotqa ✅https://t.me/deep_learning_proj

با عرض سلام دوستان اين گروه ها كامل پرشده بجز بخش طبقه بندي پزشكي كه نفر دوم و پنجم يه گروه جا هست .

Forecasting in Economics, Business, Finance and Beyond 📚 Book @Machine_learn

https://www.sciencedirect.com/journal/knowledge-based-systems ✅if=7.2 ✅@Raminmousa ✅@Machine_learn

🖥 Awesome LLM Strawberry (OpenAI o1) ▪ Github ✅@Machine_learn

Title: Computation-Efficient Neural Network Based on Model’s Saliency Performance Abstract The increasing complexity of deep neural networks has resulted in significant computational overhead, limiting their deployment in real-time and resource-constrained environments. While model pruning and quantization have been explored extensively, they often do not consider the model's saliency performance, which reflects how critical specific neurons or layers are to the overall task. This paper presents a Computation-Efficient Neural Network framework that uses model saliency to identify and preserve the most critical components of the network while reducing the computational cost by pruning less significant elements. The approach computes the saliency score of each layer or neuron, evaluates its contribution to the model's performance, and prunes the less salient parts without significant accuracy loss. By focusing on saliency, this method maintains robust performance while reducing both memory and computational demands. Experiments on image classification tasks demonstrate the effectiveness of this saliency-based pruning in achieving high efficiency with minimal performance degradation. Keyword: Deep Learning Model Compression, Convolutional Neural Networks Medical Image Classification, Quantization-Aware Training, Computational Efficiency * Submission: Nature Springer ** This paper is written by two PhD students from top universities in the USA. *** A one-page summary is attached. @reza_alvandi

فقط نفر ٤ مونده

نفر دوم واگذار شده...!

با عرض سلام موضوع سری زمانی که می خواهیم کار کنیم. پیش بینی عدم نقد شوندگی crypto market هستش. برای این منظور ۵ ارز Btc, ETH, LTc,Doge, and BNB رو در نظر داریم. مقالات مرتبط ۱-https://edoc.hu-berlin.de/handle/18452/29340 2-https://link.springer.com/article/10.1007/s00521-020-05129-6 3-https://www.researchgate.net/publication/382148896_Forecasting_of_Bitcoin_illiquidity_using_high-dimensional_features_and_hybrid_CNNRNN_Models 4-https://www.researchgate.net/publication/380570068_Forecasting_of_Bitcoin_Prices_Using_Hashrate_Features_Wavelet_and_Deep_Stacking_approach هزینه های اجاره ی سرور، جمع اوری داده ها و کدنویسی برای نفر دوم ۱۵ میلیون، نفر سوم ۱۲ و نفر ۴ ام ۱۰ میافته. هر شخص هم تسک مربوط به خودش رو باید انجام بده. کسی نیاز داشت به بنده پیام بده @Raminmousa

🖥 An Introduction to Tensors for Students of Physics and Engineering ▪ Book @Machine_learn

Hands-On Large Language Models 📚 Github @Machine_learn

🖥 UNet 3+ Implementation in TensorFlow This article presents an implementation of the UNet 3+ architecture using TensorFlow. UNet 3+ extends the classic UNet and UNet++ architecture. This article looks at each block of the UNet 3+ architecture and explains how they work and what helps improve the performance of the model. Understanding these blocks will help us understand the mechanisms behind UNet 3+ and how it effectively tackles tasks such as image segmentation or other pixel-wise prediction tasks. https://idiotdeveloper.com/unet-3-plus-implementation-in-tensorflow/ @Machine_learn