#machinelearning — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #machinelearning, aggregated by home.social.
-
Let’s take a deeper look into some of the key emerging use cases that underline the revolutionary potential that ML holds in enhancing collaboration. https://hackernoon.com/how-ml-is-transforming-the-way-offices-communicate-and-work-with-one-another #machinelearning
-
От LLM-портала до фабрики внутренних агентов: как в Авито строят корпоративного ассистента «Виталик»
По данным McKinsey, к концу 2025 года 71% компаний пробовали интегрировать искусственный интеллект как минимум в один из бизнес-процессов. Звучит мощно, но по отчётам MIT из всех пилотов в реальный продакшн переезжают около 5%. Разрыв между демо и реальностью заключается в том, что продакшн-готовое решение должно отвечать на сложные вопросы: - как из множества разносторонних источников выбрать ту информацию, которая приведёт к успешному выполнению процесса? - как обогатить языковую модель инструментами, при помощи которых она может довести процесс до конца? - как оценить качество траектории, по которой движется модель, пока решает задачу? Мы нашли ответы на эти вопросы с помощью системы, которая постепенно превращает внутренние знания и процессы Авито в корпоративных агентов. Расскажу про неё подробнее. Привет! Меня зовут Никита, я старший DS-инженер в Авито . В этой статье расскажу, как мы строили «Виталика» — нашу корпоративную экосистему ИИ-агентов. Отвечу на три главных вопроса: как процесс становится агентом, как сделать систему production-ready и куда развивать её дальше. Статья будет полезна ML-инженерам, AI-архитекторам и техническим лидам, которые строят агентные системы на базе LLM в корпоративной среде.
https://habr.com/ru/companies/avito/articles/1071126/
#ai #llm #agent #automation #claude #machinelearning #aiагенты #оркестрация_агентов #векторный_поиск #промптинжиниринг
-
Efficient Neural-Network-Based High-Resolution Radiative Transfer for CO2 Retrieval, and Application to Interferometric Sensing
Source: arXiv cs.LG
https://arxiv.org/abs/2608.14645 -
A study tests uncertainty fusion pipelines on 1,000 European Court of Human Rights cases to predict Convention violations, finding no improvement in discrimination over raw LLMs or a term-frequency baseline.
Source: arXiv cs.LG
https://arxiv.org/abs/2608.14617 -
iFuzz-Meta: An Interpretable Fuzzy Learning Framework Bridging Top-Down and Bottom-Up Knowledge Integration
Source: arXiv cs.LG
https://arxiv.org/abs/2608.14646 -
El Machine Learning no es una línea recta, es un ciclo. Desde la recolección de datos hasta el despliegue, cada paso importa. ¿Resultados OK?
Si no, toca iterar. Validación, pruebas con datos nuevos y luego al mundo real.
Así se construye IA que funciona.
Guarda esto si estás aprendiendo ML. 🚀
#MachineLearning #FlujoDeTrabajo #IA #DataScience #AprendizajeAutomático #MLtips -
Multimodal agents are building entire 3D worlds from scratch. VibeWorlding turned text prompts into open worlds—end-to-end, with 20 upvotes on Hugging Face. The paper:
-
A new diagnostic method could allow electricity substations to detect hidden insulation faults in relay-protection circuits earlier than conventional testing.
Source: Phys.org Technology
https://techxplore.com/news/2026-08-combining-electrical-faults-earlier.html -
Combining any model with GARCH(1,1) for probabilistic stock forecasting
https://thierrymoudiki.github.io/blog/2025/09/23/r/ANYMODEL-plus-GARCH -
MIT researchers have engineered bacteria that can function as transistors, allowing the team to create living circuit boards that can be printed onto a growth medium in a Petri dish.
Source: Phys.org Technology
https://techxplore.com/news/2026-08-bacteria-transistors.html -
UCLA researchers have developed a deep learning framework that reconstructs high-resolution fluorescence lifetime images from data acquired at up to five times lower spatial resolution.
Source: Phys.org Technology
https://techxplore.com/news/2026-08-ai-framework-reconstructs-high-resolution.html -
Generalist uses human demonstration data for robot learning
Source: The Robot Report
https://www.therobotreport.com/how-generalist-uses-human-demonstration-data-robot-learning/ -
🔥 Calgary Urban Heat: How the Solid-to-Tree Ratio drives an 8°C gap
My Summer 2025 ML satellite analysis reveals a direct relationship between a community's built-to-canopy footprint and surface temperature (LST):
🌳 Cooling Refuges (~29.5–31°C): Roxboro, Rideau Park, Discovery Ridge, Eau Claire (mature canopies + river valleys).
🔥 Northeast Heat Dome (~37–38°C): Marlborough, Rundle, Temple (dense low-rise footprint, minimal mature canopy).
🏗️ New Suburbs: Seton, Redstone, Rangeview (canopy lag: fully built out, but young saplings need years to mature).🌲 Urban trees are not decorative landscaping — they are critical municipal climate infrastructure.
#GIS #RemoteSensing #YYC #EarthObservation #Calgary #SpatialDataScience #LandCover #MachineLearning #Geoscience #GreennessOfCalgary #Rstats #UrbanTrees #TreeCanopy #UrbanHealth
-
FED-POSTER: THE 500+ PLATFORM PUBLISHING ENGINE NOW INTEGRATES GOOGLE CALENDAR
FED-POSTER: THE 500+ PLATFORM PUBLISHING ENGINE NOW INTEGRATES GOOGLE CALENDAR -
Roboflow Playground: Try and Compare 30 Computer Vision Models
https://blog.roboflow.com/roboflow-playground/
Comments: https://news.ycombinator.com/item?id=49335517
#HackerNews #Roboflow #Playground #ComputerVision #AIModels #MachineLearning #TechInnovation
-
OpenAI dissolves Preparedness team for AI risks
OpenAI disbanded its Preparedness team, tasked with assessing AI risks. Responsibilities distributed among existing teams.
#Anthropic #ChatGPT #IT #KünstlicheIntelligenz #MachineLearning #OpenAI #Spracherkennung #news
-
📢 Upcoming webinar — MolmoMotion: Forecasting Point Trajectories in 3D with Language Instruction
Guest: Jason Ren, Ai2 (Allen Institute for AI)
🗓 8/20 @ 9am Pacific
🔗 Register: https://opencv.org/molmomotion-forecasting-point-trajectories-in-3d-with-language-instruction-opencv-live-221/Give the model an instruction in plain language and it forecasts where points travel through 3D space, building on the pointing and tracking that the Molmo family is known for.
More on Molmo: https://allenai.org/molmo
-
Season 1 Lesson 37 Part 9 - Your First Steps in Python Python Inheritance Initialization Flow #pythonprogramming #python #dataengineer #vibecoding #softwaredeveloper #datascience #machinelearning #azure #jupyternotebook #codingtutorial #softwarengineer #dataanalysis #gcp #pythoncode #aws #learncoding
-
Stili jazzistici analizzati con l'intelligenza artificiale
https://huwcheston.github.io/ImprovID-app/index.html?noIntro=true
-
OpenAI löst Preparedness-Team für KI-Risiken auf
OpenAI hat sein Preparedness-Team aufgelöst, das KI-Modelle auf katastrophale Risiken prüfen sollte. Die Aufgaben werden auf bestehende Teams verteilt.
#Anthropic #ChatGPT #IT #KünstlicheIntelligenz #MachineLearning #OpenAI #Spracherkennung #news
-
Cheston et al. develop a machine learning pipeline that identifies 20 iconic jazz pianists from audio recordings with up to 94% accuracy
Source: Nature Machine Intelligence
https://www.nature.com/articles/s42256-026-01279-9 -
Researchers in the Riccio College of Engineering at the University of Massachusetts Amherst have demonstrated that redesigning both hardware and algorithms can make AI applications on edge devices more efficient.
Source: Phys.org Technology
https://techxplore.com/news/2026-08-edge-ai-efficient-redesigning-algorithm.html -
Bayesian Optimization with GPopt Part 2 (save and resume)
https://thierrymoudiki.github.io/blog/2021/04/30/python/misc/gpopt