23171
Admin: @Raminmousa ID: @Machine_learn link: https://t.me/Machine_learn
🔥 OmniReasoning: Pushing the Limits of Audio-Visual Joint Reasoning
💡 Summary
🔗 Links:
GitHub: https://github.com/huggingface
arXiv: https://arxiv.org/abs/2609.39490
PDF: https://arxiv.org/pdf/2609.39490
Project Page: https://pku-value-lab.github.io/OmniReasoning-Homepage/
📊 Datasets citing this paper:
https://huggingface.co/datasets/mjuicem/OmniReasoning-131K
https://huggingface.co/datasets/mjuicem/OmniReasoningBench
@Machine_learn
با عرض سلام این مقاله نفرات ۲ و ۳ رو میخواییم اسم بنویسم. هزینه به ترتیب ۳۰۰ و ۲۰۰ دلار. اگه کسی نیاز داشت به ایدیم پیام بده.
@Raminmousa1
🔥 Prism: Dynamic Sparse Attention for Native 2K Joint Video-Audio Generation Model Training
🔗 Links:
GitHub: https://github.com/huggingface
arXiv: https://arxiv.org/abs/2610.05416
PDF: https://arxiv.org/pdf/2610.05416
Project Page: https://francis-rings.github.io/Prism/
🤖 Models citing this paper:
https://huggingface.co/FrancisRing/Prism
🚀 Spaces citing this paper:
https://huggingface.co/spaces/bibibi12345/Prism
https://huggingface.co/spaces/tardellirs/model-pulse
@Machine_learn
High-Impact Research Paper – MedicalRec / GROKRec
Title: MedicalRec
Price: $1,000 USD
Looking for a ready-to-publish, novel research paper in medical AI and sustainable deep learning?
This paper introduces GROKRec, an innovative recommender framework that uses embedding vectors from the Grok language model combined with numerical features to recommend the best deep learning model for any medical image classification task — without the need to train dozens of models on the target dataset.
Key Highlights:
Addresses major real-world problems: high computational cost, energy consumption, carbon emissions, and e-waste caused by training large DL models.
Built on a newly curated public dataset MedicalRec-Bench II containing 3,500 research papers and over 6,000 model evaluation records across diverse medical imaging tasks.
Evaluated under four feature configurations (MedicalRec I) using 13 different models.
Achieves strong performance with HitRate@100 ranging from 72.43% to 77.08% — the highest among compared approaches.
Uses composite loss functions and regularization techniques for accurate recommendations.
Fully eliminates the trial-and-error process of training multiple models, significantly reducing carbon footprint.
This is a complete, self-contained research contribution with a novel dataset and a practical, environmentally conscious solution for the medical AI community.
Ideal for: Researchers, academic publishers, journals, or institutions looking for high-quality, ready-to-use work in medical image analysis, recommender systems, and green AI.
Price: $1,000 USD (one-time transfer of ownership/rights as agreed).
Interested? Contact me for the full manuscript, dataset details, or to discuss terms.
@Raminmousa1
🔥 TAPNext: Tracking Any Point (TAP) as Next Token Prediction
🔗 Links:
GitHub: https://github.com/huggingface
arXiv: https://arxiv.org/abs/2504.05579
PDF: https://arxiv.org/pdf/2504.05579
Project Page: https://tap-next.github.io/
@machine_learn
با عرض سلام این مقاله به صورت کامل واگذار میشه به همراه پیاده سازی, مجموعه داده و قالب latex. هزینه کار ۱۰۰۰ دلار
@Raminmousa1
Uniface
Automate face detection, recognition, and analysis of key facial landmarks with the Uniface Python library.
https://github.com/yakhyo/uniface
@Machine_learn
ReCamMaster: Camera-Controlled Generative Rendering from A Single Video
Source code: https://github.com/KlingAIResearch/ReCamMaster
@Machine_learn
This repository contains Jupyter notebooks for the O'Reilly book "Transformers: The Definitive Guide."
It includes code for computer vision tasks, time series analysis, audio processing, and reinforcement learning.
https://github.com/Nicolepcx/transformers-the-definitive-guide
@Machine_learn
🔖Computer Science Fundamentals from MIT
We found the textbook Mathematics for Computer Science – covering the mathematics that underlies algorithms and computer science.
Logic, graphs, combinatorics, probability, induction, recurrence relations, and discrete structures – all in one place.
⛓️ Link to the textbook
https://ocw.mit.edu/courses/6-042j-mathematics-for-computer-science-spring-2015/mit6_042js15_textbook.pdf
@Machine_learn
🔖 ML algorithms in visualizations
A useful repository that helps you understand how machine learning algorithms work – through interactive diagrams and step-by-step explanations.
You can run it in your browser or locally using Docker.
⛓ Link to GitHub
https://github.com/gavinkhung/machine-learning-visualized
@Machine_learn
🔥 8 skills = 8 free certifications >>>
AI (Microsoft) -
https://learn.microsoft.com/en-us/training/paths/get-started-artificial-intelligence/
Deep learning (NVIDIA) -
https://learn.nvidia.com/en-us/training/self-paced-courses
Data science (IBM) -
https://skillsbuild.org/students/course-catalog/data-science
Data Analyst (Microsoft) -
https://learn.microsoft.com/en-us/training/paths/data-analytics-microsoft/
Python (Microsoft) -
https://learn.microsoft.com/en-us/shows/intro-to-python-development/
SQL (Infosys) -
https://www.coursejoiner.com/freeonlinecourses/infosys-free-certification-course-9/
Java (Infosys) -
https://www.coursejoiner.com/uncategorized/infosys-launched-free-java-certification-course/
Cloud computing (AWS) -
https://explore.skillbuilder.aws/learn/course/134/aws-cloud-practitioner-essentials
@Machine_learn
با عرض سلام در مقاله زیر جهت سابمیت نیاز به نفر دوم داریم
Title: FFChurn: Fusion Former for Customer Churn Classification Based on Transformer, FEDformer, and Informer
Abstract: Customer churn prediction is a key issue in customer relationship management that directly impacts organizational profitability and has created challenges for researchers and organizations. Machine learning (ML), Ensemble Learning (EL), and Deep Learning (DL) models have achieved comparable results on this problem. In this study, Fusion Former was introduced, integrating the FEDformer, Informer, and Transformer architectures to simultaneously extract local features and long-term dependencies. The pipeline for this model includes denoising with a wavelet transform, Min-Max normalization, and hybrid adaptive feature selection based on mutual information (MI), recursive feature elimination (RFE), and the Boruta algorithm. Four different versions of the model, including binary and ternary object fusion, were evaluated on two datasets. The results showed that the Fusion Former (FED+INF+Transformer) model, with F1 scores of 0.9876 on the Dataset 1 and 0.8887 on the Dataset 2, outperformed classical machine learning models, multilayer neural networks, and other binary combinations. Also, the sensitivity analysis of hyperparameters, which included changes in the cost function, batch size, and dropout size, methods for dealing with data imbalance, which included Smote, TMG-GAN, Ib-gan, T-SMOTE approaches, and the effect of feature selection, which included four methods: MI, RFE, Boruta, and Adaptive FS (Boruta+MI+RFE), confirmed the superiority and relative stability of the proposed model.
Price:250$
@Raminmousa1
@Machine_learn
🔖 Learning Data Science through interactive examples
One of the most useful repositories for those who want to better understand machine learning.
It transforms complex concepts into visual experiments: you can study models, change parameters, and immediately see the results.
⛓ Link to GitHub
https://github.com/GeostatsGuy/DataScienceInteractivePython
@Machine_learn
با عرض سلام این مقاله نفرات ۲ و ۳ رو میخواییم اسم بنویسم. هزینه به ترتیب ۳۰۰ و ۲۰۰ دلار. اگه کسی نیاز داشت به ایدیم پیام بده.
@Raminmousa1
A Comparative Survey on Large Language Models for Biological Data
Ramin Mousa,
Ali Sarabadani
Tania Taami
Amir Ali Bengari
Omid Eslamifar
Mohammad Alijanpour Shalmani
Ehsan Karimi Shahmarvandi
Correspondence: raminmousa@znu.ac.ir
@Machine_learn
🔥 Spatial Memory Intelligence: Endowing World Models with Understanding-Driven Long-Term Memory
🔗 Links:
GitHub: https://github.com/huggingface
arXiv: https://arxiv.org/abs/2610.02521
PDF: https://arxiv.org/pdf/2610.02521
Project Page: https://spatial-memory-intelligence.github.io/
@Machine_learn
🔥 FrameMorrow: Future-guided Frame Selection with Prospective Tokens for Long-Horizon Video Generation
🔗 Links:
GitHub: https://github.com/huggingface
arXiv: https://arxiv.org/abs/2609.38839
PDF: https://arxiv.org/pdf/2609.38839
Project Page: https://yinbo0927.github.io/FrameMorrow/
@Machine_learn
📖 "A Little Book on the Fundamentals of Generative AI" - an intuitive introduction to the mathematics:
arxiv.org/pdf/2605.29713
#GenerativeAI #Mathematics #DeepLearning #AIResearch #MachineLearning #arXiv
✨ Join Best TG Channels /channel/addlist/0f6vfFbEMdAwODBk
@Machine_learn
Understanding Attention
From Q, K, V to Modern Transformer Attention
https://drive.google.com/file/d/1fCHQ5xCQJ6jZszAYf-qP3VIySbzFIEDv/view
@Machine_learn
High-Impact Research Paper – MedicalRec / GROKRec
Title: MedicalRec
Price: $1,000 USD
Looking for a ready-to-publish, novel research paper in medical AI and sustainable deep learning?
This paper introduces GROKRec, an innovative recommender framework that uses embedding vectors from the Grok language model combined with numerical features to recommend the best deep learning model for any medical image classification task — without the need to train dozens of models on the target dataset.
Key Highlights:
Addresses major real-world problems: high computational cost, energy consumption, carbon emissions, and e-waste caused by training large DL models.
Built on a newly curated public dataset MedicalRec-Bench II containing 3,500 research papers and over 6,000 model evaluation records across diverse medical imaging tasks.
Evaluated under four feature configurations (MedicalRec I) using 13 different models.
Achieves strong performance with HitRate@100 ranging from 72.43% to 77.08% — the highest among compared approaches.
Uses composite loss functions and regularization techniques for accurate recommendations.
Fully eliminates the trial-and-error process of training multiple models, significantly reducing carbon footprint.
This is a complete, self-contained research contribution with a novel dataset and a practical, environmentally conscious solution for the medical AI community.
Ideal for: Researchers, academic publishers, journals, or institutions looking for high-quality, ready-to-use work in medical image analysis, recommender systems, and green AI.
Price: $1,000 USD (one-time transfer of ownership/rights as agreed).
Interested? Contact me for the full manuscript, dataset details, or to discuss terms.
@Raminmousa1
❇️ MindDrive: A Vision-Language-Action Model for Autonomous Driving via Online Reinforcement Learning 🔥
Source code: https://github.com/xiaomi-mlab/Minddrive
@Machine_learn
فقط نفر دوم از این مقاله باقی مونده...!
Читать полностью…
با عرض سلام برای یکی از مقالاتمون تحت عنون زیر نیازمند نفر دوم و سوم هستیم.
Price: 2 --> 200$
Price 3--> 150$
Title:Skin cancer diagnosis (scd) using efficientnet-wavelet and Optimization algortithms
@Raminmousa1
Matrix Calculus for Machine Learning and Beyond! — a free ebook from MIT.
This is the 2025 MIT textbook by Alan Edelman, Steven G. Johnson, and Paige Bright.
The book directly connects matrix calculus to modern machine learning.
Inside:
* Derivatives of matrices and vectors
* Jacobian and Hessian
* Matrix decompositions
* Optimization
* Differentiation in reverse mode
* Backpropagation of error
* Automatic differentiation
* Derivatives through ODEs
* Problems focused on machine learning
This is a comprehensive mathematical bridge between linear algebra, calculus, optimization, backpropagation, and machine learning.
Free ebook:
https://geni.us/Matrix-Calculus-Book
@Machine_learn
"The Mathematics of Bitcoin" is a concise work that analyzes Bitcoin from a mathematical perspective. 📊
It utilizes probability theory, stochastic processes, martingales, combinatorics, and special functions to explore the mechanisms of the Bitcoin protocol. 🧮
In particular, the authors examine the probability of double-spending, the profitability of mining, block generation, miner strategies, and the resilience of the protocol. ⛏️
If you want to delve deeper, I also recommend "Bitcoin and Cryptocurrency Technologies" from Princeton University. This is a much broader introduction to cryptographic hash functions, digital signatures, consensus, Proof of Work, mining, transactions, anonymity, security, and the incentive system in cryptocurrencies. 🎓
The Mathematics of Bitcoin:
https://arxiv.org/pdf/2003.00001
Bitcoin and Cryptocurrency Technologies:
https://d28rh4a8wq0iu5.cloudfront.net/bitcointech/readings/princeton_bitcoin_book.pdf
@Machine_learn
اخرین زمان سابمیت این مقاله امشب...!
@Raminmousa1
هر هفته با یک موضوع تحقیقی
موضوع :تولید داده های سری زمانی با استفاده از شبکه های عصبی تخاصمی در شبکه های هوشمند
#Thesis #proposed_research
@Raminmousa1
@Machine_learn
Attention Heatmap vs Token Pruning 🔍✂️
🔗 More: https://www.overshoot.ai/blogs/an-introduction-to-token-pruning-for-vlms
#AI #MachineLearning #TokenPruning #DeepLearning #TechNews #VLM
@Machine_learn