Redução de Dimensionalidade com Grafos Ajustados Localmente
LocalMAP é um novo algoritmo de redução de dimensionalidade que ajusta dinamicamente o grafo de similaridade dos dados, tornando os ‘clusters’ mais separáveis e confiáveis, especialmente em grandes conjuntos de dados, superando limitações de métodos tradicionais.
📎https://arxiv.org/pdf/2412.15426
👨💻 https://github.com/williamsyy/LocalMAP
#LocalMAP #Clusters #MachineLearning #DataScience #DataViz #BigData #ML #GraphLearning #Arxiv #OpenSource #Python
Interesting paper on clustering and minimum spanning trees in the presence of weak oracles (~ noisy data). I quite like the MST approximation within a constant bound, as long as your noisy distance is a metric. PMLR 247:498-550, 2024. 'Metric Clustering and MST with Strong and Weak Distance Oracles'
https://proceedings.mlr.press/v247/bateni24a/bateni24a.pdf #scientificcomputing #graph #graphlearning #clustering #geometry #network
PhD Position - Neuro-Symbolic AI for Scene Understanding in Autonomous Driving
Company Description: At Bosch, we shape the future by inventing high-quality technologies and services that spark enthusiasm and enrich people’s lives. Our promise to our associates is rock-solid: we grow together, we enjoy our work, and we inspire each other. Join in and feel the difference.The Robert Bosch GmbH is looking forward to your application!
Job Description: State-of-the-art machine Learning algorithms used in domains such as autonomous driving or production systems require large amounts of training data to learn reliable models. The cost of obtaining the required training data for all relevant driving scenarios and different system configurations explodes quickly. In this context, your studies will be focused on investigating and developing methods and techniques that are capable of leveraging a-priori knowledge of the driving contexts and systems to facilitate the transfer (re-use) of already trained models for new driving contexts and evolving system configurations.Your tasks:* Survey and analyse the state-of-the-art on knowledge representation and knowledge-driven machine learning
* Research and development of novel methods / frameworks enabling the use of a-priori knowledge in existing machine learning pipelines with the goal to increase learning efficiency, performance and robustness
* Prototypical implementation and evaluation of the developed methods in concrete application scenarios (e.g. autonomous driving, production systems)
* Collaborate with Bosch research groups and relevant external research partners
* Publish the research results in academic conferences and journals
Qualifications: Education: excellent Master of Science degree in Computer Science or Mathematics, Physics, Engineering or a related subjectPersonality: positive team player who is highly motivated, has an high degree of personal responsibility, commitment and out-of-the-box thinkingWorking practice: Ideally, you are familiar with the recent developments in the field, also you want to advance the state of the art and develop new ideas, and contribute to a multi-national cutting edge project.Experience and knowledge: strong experience with machine learning, familiar with W3C Semantic technologies and Ontologies (i.e. RDF, RDFS, OWL) as well as Knowledge Graphs, strong programming skills in Python and proven experience with machine learning frameworks like Pytorch and Tensorflow and preferably have already worked on GPU ClustersLanguages: fluent in English
Additional Information: Bosch Research
Bosch Center for Artificial Intelligence (BCAI)Please submit all relevant documents (incl. motivation letter, curriculum vitae, and certificates).Apply now in just 3 minutes!
Need support during your application?
Kevin Heiner (Human Resources)
+49 711 811 12223Need further information about the job?
Adrian Trachte (Functional Department)
+49 711 811 49397
Looking for inspiration? Let yourself be inspired by the stunning and evocative Bertinoro. The best frame for a summer school maybe for this slide we can also find a better pic of the terrace?...
Applications for the International #SemanticWeb Research Summer School are open.
https://2023.semanticwebschool.org/apply/
#knowledgegraphs #ai #graphlearning #creativeAI #Bertinoro #nfdi #digitalhumanities
Apply -International Semantic Web Summer School 2023
How to Apply 1. Fill out the application form: Apply 2. Send your CV via Email to [email protected] Application deadline is March 30th April 10th, 2023. Fees and Grants To participate in the regular selection Read more…
Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges
The last decade has witnessed an experimental revolution in data science and machine learning, epitomised by deep learning methods. Indeed, many high-dimensional learning tasks previously thought to be beyond reach -- such as computer vision, playing Go, or protein folding -- are in fact feasible with appropriate computational scale. Remarkably, the essence of deep learning is built from two simple algorithmic principles: first, the notion of representation or feature learning, whereby adapted, often hierarchical, features capture the appropriate notion of regularity for each task, and second, learning by local gradient-descent type methods, typically implemented as backpropagation.
While learning generic functions in high dimensions is a cursed estimation problem, most tasks of interest are not generic, and come with essential pre-defined regularities arising from the underlying low-dimensionality and structure of the physical world. This text is concerned with exposing these regularities through unified geometric principles that can be applied throughout a wide spectrum of applications.
Such a 'geometric unification' endeavour, in the spirit of Felix Klein's Erlangen Program, serves a dual purpose: on one hand, it provides a common mathematical framework to study the most successful neural network architectures, such as CNNs, RNNs, GNNs, and Transformers. On the other hand, it gives a constructive procedure to incorporate prior physical knowledge into neural architectures and provide principled way to build future architectures yet to be invented.
Tiago F. R. Ribeiro
Ben Cardoen
Fenjan🦖
Harald Sack
Rαhul Bhαdαni