Login

Recent searches

No recent searches

Search options

Only available when logged in.
fosstodon.org is one of the many independent Mastodon servers you can use to participate in the fediverse.
Fosstodon is an invite only Mastodon instance that is open to those who are interested in technology; particularly free & open source software. If you wish to join, contact us for an invite.

Administered by:

Server stats:

10K
active users

fosstodon.org: AboutStatusProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.3.7

#machinelearning

139 posts111 participants13 posts today
Public
ResearchBuzz: Firehose

EPFL: Small- model approach could be more effective. “Small language models are more reliable and secure than their large counterparts, primarily because they draw information from a circumscribed dataset. Expect to see more chatbots running on these slimmed-down alternatives in the coming months.”

https://rbfirehose.com/2025/04/16/epfl-small-model-approach-could-be-more-effective/

ResearchBuzz: Firehose | Individual posts from ResearchBuzz · EPFL: Small- model approach could be more effective | ResearchBuzz: Firehose
More from ResearchBuzz: Firehose
#ai#AITraining#aiassisted
Continued thread
Public
Mx Verda

Also: can we have an AI Machine Learning LLM whatever thing trained ONLY on consentually offered materials, text and art?
The bare minimum effort of "you must send your materials to us" rather than trawling the internet to steal whatever isn't nailed down.
Sure, spam bots and nuisance or malicious submitters, but most won't be. Could always have a holding pen area of like 2 or 3 months before adding to training data.

Proactively request opt-outs as well, to make filtering a tiny bit easier?

#AI#OpenAI#ChatGPT
Public
WetHat💦

A Field Guide to Rapidly Improving AI Products – O’Reilly

This article subverts traditional tools-centric AI development by revealing how a focus on qualitative error analysis can uncover actionable, domain-specific weaknesses.

Its analysis, addresses both strategic and operational challenges while acknowledging the evolution of evaluation criteria in AI systems.

oreilly.com/radar/a-field-guid

O’Reilly Media · A Field Guide to Rapidly Improving AI ProductsEvaluation Methods, Data-Driven Improvement, and Experimentation Techniques from 30+ Production Implementations
#AI#MachineLearning#PromptEngineering
Public
Tuvoc Technologies

PHP and LLM! Pioneering the Future of AI-Enhanced Web Development

afriprime.net/blogs/314648/PHP

Discover how PHP and Large Language Models (LLMs) are revolutionizing web development by enabling smarter, AI-powered applications. Explore their synergy in creating dynamic, intelligent web experiences.

#PHPDevelopment
#LLM
#AIinWebDevelopment
#WebDevelopment2025
#AIPoweredApps
#FutureOfWebDev
#MachineLearning
#ArtificialIntelligence
#TechInnovation
#AIIntegration
#WebDevTrends

Public
💧🌏 Greg Cocks

Interpretable Physics-Informed Graph Neural Networks For Flood Forecasting
--
doi.org/10.1111/mice.13484 <-- shared paper
--
github.com/MehdiTaghizadehUVa/ <-- shared GitHub repository
--
#GIS #spatial #mapping #FloodForecasting #GraphNeuralNetworks #PhysicsInformedAI #Hydrology #ClimateResilience #ScientificMachineLearning #CivilEngineering #HydroGraphNet #PhyiscsNeMo #hydrology #water #flood #flooding #spatialanalysis #spatiotemporal #climatechange #model #modeling #machinelearning #AI #extremeweather #prediction #forecasting #planning #humanimpacts #loss #lossoflife #publicsafety #cost #economics #infrastructure #risk #hazard #disaster #mitigation #preparedness #naturaldisaster #HydroGraphNet #floodforecasting #dynamics #remotesensing #earthobservation

Public
Compsci Weekly

[R] Recall: A Framework for Long-Term AI Memory

zenodo.org/records/15220146

Discussions: discu.eu/q/https://zenodo.org/

ZenodoRecall: A Framework for Long-Term AI MemoryCurrent AI memory systems often suffer from excessive rigidity (as in traditional databases) or opacity(as in neural network-based approaches), hindering efficient knowledge storage, retrieval, and updating, particularly for logical or dynamic information. Recall proposes a balanced alternative: a memory framework combining a flexible graph structure with temporal awareness, logic storage capabilities, and trust-based conflict resolution. This design aims to create a more adaptable and transparent system for long-term AI memory. This paper provides a high-level introduction to the Recall framework, presenting its core concepts and design motivations while avoiding technical implementation details.
#compsci#machinelearning
Continued thread
Public
Valeriy M., PhD, MBA, CQF

#TimeSeries #Forecasting #StartupLife #DataScience #MachineLearning #SiliconValley

valeman.gumroad.com/l/Masterin

GumroadMastering Modern Time Series Forecasting : A Comprehensive Guide to Statistical, Machine Learning, and Deep Learning Models in Python📘 Mastering Modern Time Series Forecasting (preorder - release in 2025)The Definitive Guide to Statistical, Machine Learning & Deep Learning Models in PythonLet’s be honest — most forecasting books are either outdated, too shallow, or written by folks who’ve never actually built a real forecasting system. If you’ve ever felt frustrated by books that skip the basics, toss in code without explaining it, or barely touch on what forecasting really involves — you’re not alone.This is different.Mastering Modern Time Series Forecasting is your all-in-one, no-shortcuts guide to building reliable, high-impact forecasting systems. Whether you're just getting started or looking to deepen your expertise, this book takes you from rock-solid foundations to the latest advances in forecasting — including deep learning, transformers, and FTSM (Foundational Time Series Models). Written by a practitioner with over a decade of experience, who’s built production-grade forecasting systems for multibillion-dollar companies, this book is grounded in reality — not hype. The systems I’ve helped build have delivered multimillion-dollar business value, but I’ve also seen the other side: data science teams chasing shiny tools, only to ship systems that crash in production, fail silently, or burn through budgets without results. This book is a response to that — combining practical Python examples, real-world case studies, and a clear path to building forecasting solutions that actually work, scale, and deliver value. 🔍 What You'll Learn📘 Core Forecasting Foundations Grasp what forecast accuracy really means, master model validation strategies, and sidestep common pitfalls that trip up even experienced practitioners.📈 Classical Models, Done Right In-depth, modern takes on ARIMA, Exponential Smoothing, and other classical statistical and econometrics models — with clarity, not complexity.🤖 Machine Learning for Time Series Build feature-rich forecasts using state-of-the-art ML techniques that go far beyond black-box models.🧠 Deep Learning & Transformers Explore powerful deep learning architectures, including Transformer-based models — all with clear, readable PyTorch code.📊 FTSMs – Foundational Time Series ModelsExplore the rise of Foundational Time Series Models (FTSMs) — large, pre-trained models designed to generalize across domains, tasks, and time horizons. Think GPT for time series.🎯 Probabilistic & Interpretable Forecasting Move beyond point forecasts with uncertainty quantification, conformal prediction, SHAP, attention mechanisms, and explainability tools.📊 Real-World Case Studies Apply what you’ve learned on practical datasets across domains like retail, energy, and finance.🚀 MLOps & Deployment Learn how to deploy, monitor, and scale your forecasting pipelines in the real world — without the headaches.👥 Who It’s For Data Scientists & ML EngineersSolving real-world forecasting challenges and building production-ready systems. Analysts & DevelopersLooking for a practical, hands-on reference that covers both fundamentals and advanced techniques. Students, Educators & ResearchersIn need of a modern, curriculum-friendly resource grounded in both theory and application. Demand Planners & Business StrategistsFocused on delivering real value through accurate, actionable forecasts. 🧠 Why This Book Stands Out 🔍 Starts with what matters — metrics and validationBefore jumping into models, you’ll learn how to evaluate them properly so you’re building on a solid foundation. 🧠 Focuses on understanding, not just codingLearn how methods work, why they work, and when to use them — not just how to run the code. 💻 Fully documented, transparent codeNo black boxes. Every example is clearly explained so you can learn and adapt, not guess. 🔄 Updated continuously with reader feedbackBuy once, benefit forever — you’ll get lifetime updates as the field evolves. 📚 Everything in one placeFrom classical models to deep learning and FTSMs — no need to juggle multiple resources ever again. 📦 What You Get Instant download of the full book All code examples, datasets, and notebooks Free lifetime updates (including new chapters, errata fixes, and bonus content) Exclusive early access to upcoming bonus chapters & Q&A sessions 💸 Pricing 🎉 Introductory Launch Price Suggested: $29 | Minimum: $20 This is the initial price — it will increase as more chapters, tools, and content are released. If you find value or want to support the project, feel free to pay what it’s worth to you ❤️ Ready to take your forecasting skills from stats to neural nets, and from theory to real-world deployment?👉 Hit “Buy Now” and start mastering forecasting like never before.
Public
Habr

[Перевод] Рекомендательная система для вашего каталога научных работ (и не только!)

Привет, Хабр! Как выжать максимум из своего архива документов? В мире, где объем информации стремительно растет, найти релевантные материалы среди собственных файлов — задача не из простых. В этой статье мы расскажем, как с помощью инструментов обработки естественного языка и теории графов создать умную рекомендательную систему, которая поможет находить нужные документы: будь это научные статьи, презентации или таблицы с экспериментами и даже если они хранятся в самых разных форматах.

habr.com/ru/companies/otus/art

ХабрРекомендательная система для вашего каталога научных работ (и не только!)Используем обработку естественного языка и теорию графов для сравнения и рекомендации различных типов документов. Введение Почти все проекты начинаются с одного важного этапа — активных...
#python#nlp#ml
ExploreLive feeds
About