Sign in to view Maria Fernanda’s full profile
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Campinas, São Paulo, Brazil
Sign in to view Maria Fernanda’s full profile
Join with email
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
464 followers
393 connections
Sign in to view Maria Fernanda’s full profile
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Maria Fernanda
Join with email
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Maria Fernanda
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Sign in to view Maria Fernanda’s full profile
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
About
Welcome back
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
New to LinkedIn? Join now
Activity
464 followers
-
Maria Fernanda Rodriguez shared thisI’m happy to share that I’m starting a new position as DS Engineer at Mercado Livre Brasil #oicheguei 💛
-
Maria Fernanda Rodriguez shared thisI’m happy to share that I’ve obtained a new certification: Data Scientist from Udacity! #nanodegree #udacity #datascientist
-
Maria Fernanda Rodriguez shared thisI am pleased to share that I completed an intensive beginner-level Korean course at Kyonggi University in Suwon, South Korea. This experience allowed me to immerse myself in Korean culture and gain first-hand insight into its educational system, as well as its dedication and commitment. It was a period of significant learning and personal growth. I am grateful for this opportunity and the meaningful connections I made. 감사합니다🫰🏼 #경기대 #KGU
-
Maria Fernanda Rodriguez shared thisHaving just finished reading ‘Storytelling with Data by Cole Nussbaumer Knaflic,’ I’d like to share a few key takeaways and highlights from the book. However, to delve into the complete and detailed content, I strongly recommend reading the entire book for a more comprehensive understanding. #storytelling #datascience #datavisualization https://lnkd.in/d_EBRi4tWhy is it important to communicate data effectively?Why is it important to communicate data effectively?
-
Maria Fernanda Rodriguez shared thisStreamlit and Hugging Face are excellent choices if you need to quickly develop a prototype or web app. Both tools offer fast and efficient solutions for presenting functional results without the requirement of extensive front-end development knowledge. On one hand, Streamlit is compatible with various libraries such as TensorFlow, PyTorch, and Scikit-learn. On the other hand, Hugging Face is widely used by researchers, developers, and data scientists for building, training, and deploying NLP models. These tools are ideal for rapid prototyping, creating MVPs (Minimum Viable Products), developing applications, and sharing outcomes with customers and users. In this repository, I present a Streamlit Application integrated with a pre-trained ViT (Vision Transformer) model for digit classification. #streamlit #fastapi #huggingfaceGitHub - mafda/ml_with_fastapi_and_streamlit: Prototyping a Machine Learning Application with Streamlit, FastAPI, Hugging Face and DockerGitHub - mafda/ml_with_fastapi_and_streamlit: Prototyping a Machine Learning Application with Streamlit, FastAPI, Hugging Face and Docker
-
-
Maria Fernanda Rodriguez shared thisCertificado de conclusão do curso Mineração de Dados Complexos da Universidade Estadual de Campinas Muito feliz de mais uma etapa concluída. Obrigada aos professores e monitores do curso pelos ensinamentos. Sem dúvida uma jornada de muito aprendizagem. E meu especial agradecimento para Camilo Ariza pela motivação e apoio incondicional 💜 #unicamp #mdc
-
Maria Fernanda Rodriguez shared thisApresentação final do projeto: Analise da gravidade de artrose no joelho para o curso de Mineração de Dados Complexos da Universidade Estadual de Campinas https://lnkd.in/dT4uraYr Muito obrigada ao professor Zanoni Dias e aos monitores do curso Gabriel Bianchin de Oliveira e Lucas David pelos ensinamentos. Sem dúvida uma jornada de muito aprendizagem. E meu especial agradecimento para Camilo Ariza pelo apoio incondicional 💜 Turma MDC/013: https://bit.ly/mdc013pf Sobre o curso: https://ic.unicamp.br/~mdc #mdcunicamp #icunicamp #unicamp #datascience #cienciadedados
-
Maria Fernanda Rodriguez shared thisMuito feliz pela excelente oportunidade de conhecer de perto a Usina de Araçariguama e todo o processo de sucata. Obrigada Leonardo Magalhães dos Reis por todo o conhecimento e ao pessoal da Microsoft Brasil 💛 💙 GerdauMaria Fernanda Rodriguez shared thisA Gerdau é a maior produtora brasileira de aço e maior recicladora de sucata ferrosa da América Latina. Ontem tivemos a oportunidade de conhecer a Usina de Araçariguama para entender como funciona todo o processo da sucata. Uma experiência indescritível. Obrigada Gerdau pela oportunidade ! Eduardo T. Kodama Marcel Silva Jussara Gurgel Fonseca Robson Melo Rodrigo Marques Rafael Augusto M.
-
Maria Fernanda Rodriguez reacted on thisMaria Fernanda Rodriguez reacted on thisRecentemente implementei uma melhoria significativa em um pipeline de Machine Learning em produção, tornando o processamento de requisições mais rápido, confiável e eficiente, com impacto direto em custos e performance. O sistema integra múltiplos modelos e APIs, com comunicação entre instâncias, armazenamento em buckets e execução em containers orquestrados, garantindo que cada etapa funcione de forma coordenada e escalável. Toda a solução conta com workflows de retreinamento automático, validação de versão e pipelines robustos de experimentação e deploy. Principal aprendizado: Machine Learning em produção vai além de treinar modelos, é sobre arquitetura robusta, interação entre sistemas, automação inteligente e impacto real no negócio. Seguimos evoluindo e aprimorando continuamente 🚀
-
Maria Fernanda Rodriguez liked thisMaria Fernanda Rodriguez liked thisSeguimos investindo no crescimento do MELI no Brasil! Anunciamos o aporte histórico de R$ 57 bilhões no Brasil para este ano. O valor representa um salto de 50% em relação aos R$ 38 bilhões investidos em 2025. O plano de expansão inclui a abertura de 14 novos Centros de Distribuição Fulfillment ainda este ano, atingindo a marca de 42 unidades Fulfillment no Brasil - um crescimento de 50%. Com isso, estamos também nos preparando para a abertura de cerca de 10 mil novos postos de trabalho, elevando o nosso time para mais de 70 mil pessoas até o final de 2026. Esse movimento reforça nosso compromisso com o país e com o desenvolvimento do nosso ecossistema. Seguimos empreendendo e superando limites porque sabemos que o melhor está chegando. #OrgulhoMELI
-
Maria Fernanda Rodriguez reacted on thisMaria Fernanda Rodriguez reacted on thisRecordar que cada persona tiene una historia; que altos estándares de esfuerzo y disciplina constante pueden llevarte a alcanzar tus sueños con más certeza, fueron parte de los mensajes que nos dejó este espacio tan especial de La Nota Económica la semana pasada. Gracias Liliana Sierra Cárdenas por promover momentos de apertura y discusiones profundas acerca de lo que realmente significa equidad de género, porque todo debe partir de la meritocracia y la constancia. A las crack con quienes compartimos un rato Alexandra Quiroga Ana Rocio Sabogal Henao Sandra Paola Charris Ibarra Marcela Jacome Vergel CLAUDIA VIVIANA JAIMES GONZALEZ MARINE MORE MORERA Angela Murillo Peñuela Claudia Sepulveda Andrea Marcela Cabrejo Borda #Mujeres2026
-
Maria Fernanda Rodriguez reacted on thisMaria Fernanda Rodriguez reacted on thisI am pleased to share the publication of our article, “Control System for the Navigation of the Agricultural Robots: A Review,” in the Journal of Field Robotics. This publication represents one of the outcomes of my doctoral research in Agricultural Engineering, in the area of Agricultural Machinery. The work presents a review of control systems applied to the navigation of agricultural robots, contributing to the discussion on technologies that support automation and precision in agricultural environments. The article is available as open access at the following link: https://lnkd.in/eGn63kA6 I am grateful for the opportunity to contribute to this field of research.Control System for the Navigation of the Agricultural Robots: A ReviewControl System for the Navigation of the Agricultural Robots: A Review
-
Maria Fernanda Rodriguez liked thisMaria Fernanda Rodriguez liked this🚀 ¡Arrancamos nuestro ciclo de webinars 2026 con un tema imperdible! Desde IEEE RAS Colombia Chapter te invitamos a nuestro primer webinar del año: “Cómo explorar datos RGB-D de la cámara de un iPhone” 📱🤖 Tendremos como invitado al Ph.D Olmer García Bedoya, quien compartirá su experiencia y conocimientos en un espacio ideal para aprender sobre visión por computador, sensores y aplicaciones en robótica. 📅 Sábado 28 de marzo 🕙 10:00 a. m. Una oportunidad increíble para seguir fortaleciendo nuestra comunidad, aprender de expertos y explorar nuevas herramientas tecnológicas. ✨Youtube: https://lnkd.in/epsgR5-Q Google Meet: https://lnkd.in/eUETiu5b #IEEERAS #IEEERASColombia #Robotics #Ingeniería #ieeecolombia
-
Maria Fernanda Rodriguez liked thisMaria Fernanda Rodriguez liked thisO Mercado Livre acaba de lançar a campanha "Raspe seus Dados"! Para reforçar um cuidado simples, fizemos uma ação para incentivar a remoção de informações pessoais antes do descarte de embalagens. A ideia é transformar esse cuidado em um hábito no dia a dia. Por isso, os primeiros três mil pedidos feitos a partir do dia 15 de março, na landing page da campanha que será divulgada no Instagram @mercadolivre, terão etiquetas especiais que revelam cupons exclusivos depois que os dados são removidos. Esse é mais um passo do Mercado Livre para fortalecer a segurança e a confiança em todo o nosso ecossistema. #MercadoLivre #ProteçãoDeDados #DiaDoConsumidor
-
Maria Fernanda Rodriguez reacted on thisMaria Fernanda Rodriguez reacted on thisVery happy to share that our journal cover is out 🤓! The cover artwork from our recent paper on Bayesian Optimization for Coarse-Grained Models has been featured in the Journal of Chemical Theory and Computation. Special thanks to Carlos Alberto Martins and Rodrigo A. Vargas-Hernandez for leading the work and to Aline Gonçalves for the final design of the cover artwork. 🔗 https://lnkd.in/gmwFRfSv
-
Maria Fernanda Rodriguez liked thisMaria Fernanda Rodriguez liked thisPé no acelerador O Mercado Livre acaba de dar mais um passo rumo ao topo; a empresa acaba de anunciar sua equipe do STOCK CAR. Reforçando os valores de velocidade e eficiência a Mercado Livre Racing contará com os já conhecidos pilotos: Helio Castroneves, Thiago Camilo e Cesar Ramos, que competirão em 12 etapas. Em uma entrevista ao meio e mensagem a marca reforçou que essa decisão veio de um conjunto de análises permeadas inclusive por já ser líder absoluta na venda de Veículos, Peças e Acessórios (11 milhões de anúncios ativos). Da pra imaginar o tamanho do orçamento disponível para o Mercado Livre gastar este ano? De dar inveja em qualquer um... uma coisa não podemos negar; pensa num dinheiro que ta sendo bem usado! .
-
Maria Fernanda Rodriguez liked thisMaria Fernanda Rodriguez liked thisUm dia histórico! Lançamos hoje a Mercado Livre Racing, inaugurando nossa presença na Stock Car Pro Series, a principal categoria do automobilismo brasileiro. Operada pela A.Mattheis, uma das equipes mais vitoriosas da categoria e com um trio de pilotos de altíssimo nível - Hélio Castroneves, Thiago Camilo e César Ramos, a Mercado Livre Racing representa nosso DNA de velocidade e alta performance. Além disso, é uma plataforma para reforçar nosso engajamento com as marcas de grande relevância do nosso ecossistema através do Mercado Ads Brasil. Um agradecimento especial a todas e todos que tornaram isso possível: Priscila Aguiar Rayel Desiree Caterini Thais Krauss Natália Brumati Andre Beisert ROBERTA DONATO Luiz Augusto Vergueiro + time de ADs, Marketing e Jurídico. É o Mercado Livre acelerando nas pistas e nas entregas! #OMelhorTáChegando
Experience & Education
-
Universidade Estadual de Campinas
****** ** ********** * *** ********* undefined undefined
-
-
************ ******** ** ********
******** ****** undefined
-
View Maria Fernanda’s full experience
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Welcome back
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
New to LinkedIn? Join now
View Maria Fernanda’s full profile
-
See who you know in common
-
Get introduced
-
Contact Maria Fernanda directly
Other similar profiles
-
Tiago Lôbo
Tiago Lôbo
Laboratório de Computação Científica e Visualização - LCCV
229 followersMaceió, AL
Explore more posts
-
Compilers Lab
8K followers
We are happy to share PEAK, an AI assistant that helps optimize GPU kernels, recently published on arXiv by Angelica Moreira, Ph.D. and her collaborators. Angélica is a former PhD student from UFMG's Compilers Lab and is currently a Senior Research Software Engineer at Microsoft Research Redmond. PEAK is the result of a collaboration between Microsoft Research and Stanford University. The assistant optimizes GPU kernels using natural-language instructions: performance experts describe optimization steps (like how to tile loops or use shared memory) in simple language. These steps are then applied by an AI model and automatically checked for correctness and performance. PEAK supports three GPU backends: CUDA, HIP, and HLSL. It was evaluated on matrix multiplication kernels using 16 optimization steps. The results show that PEAK can match the best performance reported for the hardware, even in cases where no vendor-provided library is available. PEAK: A Performance Engineering AI Assistant for GPU Kernels Powered by Natural Language Transformations, by Muhammad Usman Tariq (Stanford University), Abhinav Jangda (Microsoft Research Redmond), Angélica Moreira (Microsoft Research Redmond), Madan Musuvathi (Microsoft Research), and Tyler Sorensen (Microsoft Research Redmond and University of California, Santa Cruz). Link: https://lnkd.in/dJCPqErw
108
6 Comments -
Filipe Medeiros
CESAR • 2K followers
Pattern recognition in hate speech detection: a neural network and ensemble approach TL;DR: New published paper! Hate speech on online platforms is a growing problem with significant social impacts. This work proposes an approach for binary classification of hate speech in Portuguese texts using machine learning and deep learning algorithms. The experiments were conducted on an annotated dataset, with textual representations generated by pre-trained GloVe word embeddings. The voting-based model, which combines the outputs of the base classifiers, achieved the best overall performance, reaching an F1-score of 0.76. The results demonstrate the effectiveness of neural networks, especially in capturing complex textual patterns, and highlight the potential of combined approaches for the hate speech classification task. This study reinforces the importance of exploring diverse architectures and preprocessing techniques tailored to the specific characteristics of the Portuguese language. https://lnkd.in/dqwwHH_Q
51
1 Comment -
Marija Kopanja
BioSense Institute • 772 followers
🇧🇷 I'm excited to share that our paper has been accepted to the main track of the World Conference on eXplainable Artificial Intelligence 2026 in Brazil! 🇧🇷 Great work with my supervisors Luca Longo and Miloš Savić! In this work, we explored our CORTEX (Cost-Sensitive Rule and Tree Extraction) algorithm on a real-world, highly imbalanced forest cover detection problem using Sentinel-2 cloud-free imagery. 🛰 This is the first application of CORTEX in two modeling paradigms (see diagram below), as inherently interpretable (ante-hoc) classifier and as a post-hoc surrogate model for MLP network. 📊 The results show that CORTEX produces smaller and simpler rules while maintaining competitive predictive performance compared to traditional decision tree model. 👀 For more details on CORTEX, check out our paper: https://lnkd.in/gzMj9vAw We're currently preparing an open-source python package, with GitHub repo to be released soon (happy to share it upon request in the meantime). If you're interested in contributing or collaborating, feel free to reach out! 👩🏻💻 #XAI #explainability #interpretable #MachineLearning #RemoteSensing #ForestDetection #Sentinel2 #research #AI #PhD
73
4 Comments -
Jorge Trivilin
MLOps Digest • 4K followers
On a cold Saturday morning here in São Paulo, immersed in interpretability and explainability of Machine Learning models (XAI), I decided to create this repository. Born purely out of fun and educational purpose, it organizes scripts to train models (using Scikit-learn) and, more importantly, apply methods like LIME and SHAP to understand their decisions. Plus, unit testing. Although it started as a personal learning project, it can be effectively used to generate both local and global explanations for classification models all executable via the command line. :) https://lnkd.in/df2qub8A
8
-
IEEE Open Journal of Circuits and Systems (OJCAS)
1K followers
📄We are happy to share this article from Vol. 6 : "End-to-End Neural Video Compression: A Review." ✍️ Authored by: Jiovana S. Gomes; Mateus Grellert; Fábio L. L. Ramos; Sergio Bampi Video compression is crucial in providing high-quality video services under limited network and storage capacities, traditionally achieved through hybrid codecs. However, as these frameworks reach a performance bottleneck with compression gains becoming harder to achieve with conventional methods, Deep Neural Networks (DNNs) offer a promising alternative. By leveraging DNNs’ nonlinear representation capacity, these networks can enhance compression efficiency and visual quality. Neural Video Coding (NVC) has recently received significant attention, with Neural Image Coding models surpassing traditional codecs in compression ratios. Therefore, this survey explores the state-of-the-art in NVC, examining recent works, frameworks, and the potential of this innovative approach to revolutionize video compression. 🔗 Continue reading on IEEE Xplore here👉 https://loom.ly/EQ8KIrc #IEEE #CircuitsAndSystems #Research #Innovation #IEEEJournals #ElectricalEngineering
17
-
Sandro Weigert
Mondelēz International • 2K followers
Word2Vec vs. FastText: Which is better for NLP? Train from scratch or use a pre-trained model? 🤔 I put two NLP giants to the test for semantic analysis in Portuguese! 🚀 The Experiment: 1️⃣ I trained a Word2Vec model from the ground up using the entire Portuguese Wikipedia. 2️⃣ I pitted it against Facebook's massive pre-trained FastText model. 🏆 The Verdict: For general-purpose quality, speed, and robustness, pre-trained FastText is the clear winner! Its biggest advantage? It understands subwords, meaning it can generate meaningful representations even for words it's never seen before. No more "Out-of-Vocabulary" errors! Fun fact: The FastText model was so nuanced it learned that "kitten" (gatinho) is semantically similar not just to "puppy", but also to affectionate terms like "silly one" (bobinho), capturing deep cultural context. A great practical lesson in leveraging state-of-the-art models to accelerate development and achieve superior results. #NLP #DataScience #MachineLearning #Python #Word2Vec #FastText #AI #IBMEC
26
1 Comment -
Applied Sciences MDPI
12K followers
#highlycited paper 📚Impacts of #FeatureSelection on Predicting Machine Failures by Machine Learning Algorithms 🔗https://lnkd.in/gzfDvB-7 👨🔬by Francisco Elânio Bezerra, Geraldo Cardoso de Oliveira Neto, Gabriel Magalhães Cervi, Rafaella Francesconi Mazetto, Aline Mariane de Faria, Marcos Vido, Gustavo Araujo Lima, Sidnei Alves de Araújo, Mauro Sampaio and Marlene Amorim 🏫FEI University #machinelearning #machinefailure
-
Esteban Carisimo
Cloudflare • 600 followers
🚀 Borges Full Refactor: Learning AS-to-Organization Mappings with LLMs The Internet isn’t just a graph of Autonomous Systems (ASes)—it’s an ecosystem of organizations constantly reshaping through mergers, acquisitions, and rebrandings. Understanding who runs which networks is critical for management, security, and policy. That’s why we built Borges, a new approach to AS-to-Organization mapping that combines: 📑 PeeringDB metadata (org IDs, notes, aka fields) 🗂️ WHOIS-based AS2Org mappings 🌐 Web signals (redirects, domains, favicons) 🤖 LLM-powered extraction from unstructured text Now, Borges is a full Python package & CLI—complete with tests, CI, docs, and a streamlined pipeline. 🔧 What’s new in the refactor 📦 Modular src/borges/ package with a renewed CLI & new data downloaders 🧪 Full pytest suite + CI across Python 3.9–3.12, style & type checks ⚙️ Stronger WHOIS parsing, consolidation, checkpointing, and logging ⚡ 10× faster batching & rate-limiting in vision-based stages 📚 Expanded docs, hopefully easing reproducibility 💡 Why it matters Borges improves on AS2Org by handling unstructured data, multilingual text, and real-world data—capturing organizational structures at Internet scale. After improving prompt engineering, a full run costs just $3 in OpenAI's API usage. 👉 Try it out and share feedback—especially on edge cases and org-merging heuristics. 🔗 Repo: https://lnkd.in/gG_gRMtH
25
2 Comments -
Santiago DIaz, MSc
Nexia Capital • 704 followers
good post, MCP is a good approach and protocol to use different IA systems; i would like to add that we have to mitigate namespace tyosquatting, preference manipulation attack, tool poisoning, rug pulls, coss-server shadowing attack, it must be strengthen code integrity verification. All dependencies and server binaries should be build and verified thought reproducible builds, cryptographic signing and checksum validation to confirm that the deployed code matches audited sources.
1
-
Daniel Liden
Databricks • 4K followers
MLflow Ambassador Joana Ferreira just published an excellent article on building consistent evaluation pipelines for GenAI applications, and it tackles a problem most teams don't realize they have until it's too late. The core issue the article addresses is evaluation drift: when your development testing uses different standards than your production monitoring. You end up with a model that passes all your local tests but behaves differently in production, with no consistent way to compare quality across environments. Joana's solution centers on the "write once, use everywhere" principle: define your evaluation scorers once in MLflow, then apply those exact same scoring functions during local development AND for continuous production monitoring. No more wondering if your prod quality metrics actually match what you tested during development. What makes this walkthrough particularly valuable is the practical implementation. It shows you how to build custom scorers that check both output format and correctness, then demonstrates how the same evaluation logic works seamlessly across your entire pipeline. This kind of systematic approach to evaluation consistency is something more GenAI teams need to adopt. Check out the full article: https://lnkd.in/gC9rSehY #MLOps #GenAI #LLMOps #MLflow #AI
20
1 Comment -
Luis Lamb
MIT Sloan School of Management • 14K followers
https://lnkd.in/eRs2wG6x Solving the kidney exchange problem via graph neural networks with no supervision Pedro Foletto Pimenta Pedro Henrique da Costa Avelar This paper introduces a new learning-based approach for approximately solving the Kidney-Exchange Problem (KEP), an NP-hard problem on graphs. The KEP consists of, given a pool of kidney donors and patients waiting for kidney donations, optimally selecting a set of donations to optimize the quantity and quality of transplants performed while respecting a set of constraints about the arrangement of these donations. The proposed technique consists of two major steps: the first is a Graph Neural Network (GNN) trained without supervision; the second is a deterministic non-learned search heuristic that uses the output of the GNN to find a valid solution. To allow for comparisons, we also implemented and tested an exact solution method using integer programming, two greedy search heuristics without the machine learning module, and the GNN alone without a heuristic. We analyze and compare the methods and conclude that the learning-based two-stage approach is the best solution quality, outputting approximate solutions on average 1.1 times more valuable than the ones from the deterministic heuristic alone.
37
1 Comment -
Martin Ryner
Linköpings universitet • 887 followers
Transformer attention layers operate by projecting high-dimensional token representations into low-rank subspaces defined by per-head key/query/value projections. This inherently places a rank constraint on how relationships between items can be represented, and it forces the model to approximate rich interactions via combinations of rank-limited components. Our paper “Orthogonalization of data via Gromov-Wasserstein type feedback for clustering and visualization” provides a theoretical and algorithmic framework for dealing with full and low-rank structure in relational data representations by adapting transition probabilities and iteratively refining cluster orthogonality via Gromov-Wasserstein-inspired feedback. It shows how a low-rank Markov transition representation can be refined to better reflect true structure and interpretability, and how such refinements converge to meaningful solutions with spectral gap enhancement. This connects directly to issues in multi-head attention: -Both involve representing pairwise relationships (attention scores vs. transition affinities) in a full-rank or approximately low-rank form. -Low-rank factorization can produce spurious grouping or similarity structure that doesn’t reflect the true geometry, unless guided by careful feedback or regularization. -The GW-feedback orthogonalization that we exemplify with shows some ideas how to improve, stabilize, and interpret the geometry of such low-rank relational structures, offering insight into how optimization can go astray in ill-conditioned, low-rank parameter spaces, but also how the implication of the approximation can be assessed. Give it a read: Low-rank constraints are not just computational conveniences — they have real geometric and optimization implications, and without appropriate mechanisms (feedback, orthogonalization, spectral refinement), learned low-rank representations (like attention maps or transition matrices) can be misleading or unstable. I’d like to hear your thoughts on this. Is it needed? Should we care? What’s a good future architecture? What should I read? Merry Christmas!
10
-
Ben Swallow
London Mathematical Society • 1K followers
New preprint from PhD student Chenglei Hu in which we apply approximate Bayesian inference methods and extreme value theory, combined with ML feature selection methods, to forecast occurrence and severity of wildfires across Portugal https://lnkd.in/eiTghyx4
12
-
Youssef Hosni
Aalto University • 116K followers
If you want to build a multimodal RAG application, check the 𝐁𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐌𝐮𝐥𝐭𝐢𝐦𝐨𝐝𝐚𝐥 𝐑𝐀𝐆 𝐀𝐩𝐩𝐥𝐢𝐜𝐚𝐭𝐢𝐨𝐧 series of articles I have recently published on To Data & Beyond: 1️⃣ Introduction to Multimodal RAG Applications https://lnkd.in/ddkMHA84 2️⃣ Multimodal Embeddings https://lnkd.in/dYa5PC-x 3️⃣ Multimodal RAG Application Architecture https://lnkd.in/dHxjqUk3 4️⃣ Processing Videos for Multimodal RAG https://lnkd.in/dFfGj-FN 5️⃣ Multimodal Retrieval from Vector Stores https://lnkd.in/daw9d4qP 6️⃣ Large Vision Language Models (LVLMs) https://lnkd.in/dvEVwrc4 7️⃣ Multimodal RAG with Multimodal LangChain https://lnkd.in/djYPWcmY 8️⃣ Putting it All Together! Building a Multimodal RAG Application https://lnkd.in/dGhCdJGV --------------------------------------------- ➡ Do not miss the 50% discount on the To Data & Beyond yearly subscription: https://lnkd.in/dZ2Sjkhk ➡ All My 7 Books, One Button Away, With 40% Off: https://lnkd.in/dqzXcDyJ
35
-
Raphael Blankson
AI Makerspace • 3K followers
🔬 I just evaluated 11 retrieval strategies and the results shocked me. Using the Ragas framework, I explored how different retrieval methods perform in real-world RAG systems. Here’s what I found: 1️⃣ Old tech can still win. BM25 from the 1970s! was the fastest retriever (3.08s/query). Sometimes simplicity wins over sophistication, especially when speed matters more than perfect accuracy. 2️⃣ Recall > Precision. The top retrievers hit 100% recall. Missing info kills answers faster than noisy context. Your LLM can filter noise, but it can't invent missing facts. 3️⃣ Measure trade-offs, not just scores. Contextual Compression hit 100% precision/recall but cost $0.056/query vs BM25’s $0.054. At 1M queries, that is $2,300 difference. Production decisions require cost + latency + accurate data. 💭 Open Research Question: Are our synthetic evaluation datasets too friendly to the LLMs that generated them? What is the optimal synthetic-to-real ratio for reliable RAG benchmarks? 📚 Notebook: https://lnkd.in/g-uhq8Rd 🎥 Loom Video: https://lnkd.in/gmBVqwXC What's your experience with RAG evaluation? Have you found synthetic datasets reliable, or do you always validate with real user queries? Huge thanks to AI Makerspace, 👨🏫🤖 "Dr. Greg" Loughnane, Chris "The Wiz 🪄" Alexiuk, Laura Funderburk, Ovo Okpubuluku, Tyler Laughlin for inspiring this kind of thinking! #RAG #LangChain #Ragas #LLM #AIMakerspace #AIResearch #AIEngineering
21
4 Comments -
Dr. Juan Camilo Orduz
PyMC Labs • 9K followers
This weekend we merged two PyTensor Example notebooks by Ricardo Vieira and Jesse Grabowski - Intro to PyTensor: https://lnkd.in/dkEk7Frr - Normalizing Flows in PyTensor: https://lnkd.in/dzvPEE6k What is PyTensor? A library to define, manipulate, and compile computational graphs. This library is the backbone of PyMC but it has much more potential on its own beyond Bayesian modeling. Check it out!
66
1 Comment -
Joel Niklaus
Hugging Face • 7K followers
🔥 FineTranslations just dropped: 1 Trillion tokens of parallel text across 500+ languages! Created by Guilherme Penedo and team at Hugging Face, this dataset tackles a major gap: most models struggle with English→X translation, especially for low-resource languages. 💡 The insight: Instead of translating English content into other languages, they started with authentic non-English web pages from FineWeb2 and translated TO English. This creates the perfect parallel data for training better English→X models. 🎯 Key features: - 500+ languages with document-level translations - Smart chunking preserves context across sentences - Fully open & reproducible (ODC-By license) Use it for translation model fine-tuning, multilingual LLMs, or even English pre-training.
53
4 Comments -
International Conference on Evaluation and Assessment in Software Engineering (EASE)
725 followers
EASE 2025 Paper spotlight 🔦 (Full research paper) "Assessing the Bug-Proneness of Refactored Code: A Longitudinal Multi-Project Study" By Isabella Ferreira, Lawrence Arkoh, Anderson Uchôa, PhD, Ana Carla Bibiano, Alessandro Garcia & Wesley K. G. Assunção Refactoring: Fix it or break it? Developers refactor code to make it cleaner, easier to maintain, and — in theory — less buggy. But what if some refactorings are actually introducing bugs? This #EASE2025 paper dives deep into 12 popular open-source projects, analyzing over 27,000 refactorings and nearly 55,000 bugs to answer a tough question: 👉 Which types of refactoring lead to more bugs — and how soon do they strike? Key takeaways: a) Single refactorings (like renaming or extracting one method) are over 3x more likely to introduce bugs than composite ones (multiple coordinated changes). b) On average, it takes just 7 changes after a single refactor for a bug to appear. With composite refactors, it takes closer to 29. c) Even well-meaning “pure” refactorings (root-canal style) — meant just to clean up code — can still introduce bugs, especially when done alone. d) Extract Method and Inline Method were the riskiest refactoring types — often leading to bugs despite their simplicity. e)The presence of multiple code smells before refactoring significantly increases the risk of bugs later. Why this matters: a) Teams relying on frequent micro-refactors should favor composite changes and watch for silent regressions. b) Tools could be built to predict bug-prone refactors based on historical patterns. c) Code reviews and CI might need extra scrutiny on certain refactoring types (yes, even the “simple” ones!). For full access to the paper: https://lnkd.in/dEVVP9b4 🔗 Discover more gems like this at #EASE2025 → https://lnkd.in/duhcSfUp running from June 17-20 2025 Have you ever introduced a bug while “just” cleaning up code? 👇 Drop your refactoring stories in the comments — lessons welcome. #EASE2025 #Refactoring #SoftwareMaintenance #BugProneCode #CodeSmells #DeveloperExperience #EmpiricalSE #SoftwareEngineering #TechDebt #DevTools #SoftwareQuality #SoftwareReliability #QualityAssurance (QA) #CodeQuality #SoftwareEvolution #CodeMetrics #SoftwareAnalytics #LongitudinalStudy #EmpiricalResearch (reinforces #EmpiricalSE) #AcademicResearch #CleanCode #ContinuousImprovement #SoftwareArchitecture #OpenSourceSoftware (if not already included)
12
-
Alexander Thomas Savvides
McKinsey & Company • 606 followers
A recent study out of Maringá State University and Paraná Federal University introduces KAN-Mixers, a new deep learning architecture that replaces traditional convolution or attention mechanisms with Kolmogorov-Arnold Networks (KANs). By integrating KANs into MLP-based architectures, the authors propose a model that is not only more interpretable but also highly competitive in performance, achieving 0.9030 accuracy on Fashion-MNIST and 0.6980 on CIFAR-10 clearly surpassing MLP, MLP-Mixer, and standalone KAN models. The key insight: complexity in feature extraction can be modeled with greater transparency and efficiency by moving beyond convolutional priors. KAN-Mixers mark an important step toward models that are simpler, more theoretically grounded, and easier to interpret without compromising on accuracy. Is this a signal that post-convolutional architectures are maturing?
13
1 Comment
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content