Ergodic’s Post

We are not guided by expectations or beliefs in conducting our science. What we do focus on is data. In other words, reliable and methodologically acquired information based on facts. #HardData #ScientificMethod #ScientificConclusion #ScienceInformation

⁉ In the last two days, 5 exabytes of data have been generated, which is more than what was produced from the beginning of humanity until 2003❗ ❌ We are producing huge amounts of synthetic data, but what is the quality of that data? How can we better extract the knowledge from it? In this paper, we are not solving the entire grand problem, but we are proposing a strategy to optimize synthetic data set generation to allow for better #ML models for prediction of #tunnelling-induced #building #damage. You can read our article here: https://lnkd.in/e3P4sbz5 Thanks to the lead author Ali Gamra and co-author Bahman Ghiassi!

Reviewing assumptions is crucial for causal inference. Why? To make sure they're met! It's also important to determine if the assumption seems reasonable. An article I recently came across discusses this for the ICH E9 (R1) addendum on estimands. Vansteelandt & Van Lacker discuss parts of the addendum through a causal inference lens using the counterfactual approach. I work mostly with observational data. However, it can be helpful to think of things in a RCT framework then how that could apply to observational data. (Credit to Matt Tenan, for commenting this on a previous post. I find that approach quite useful) While reading this article, I tried to think of ways it could apply to observational data. Let me know in the comments your thoughts! #CausallyCurious #RealWorldEvidence #CausalInference #RealWorldData [Link: https://lnkd.in/euc23s84]

Ordering-Based Causal Discovery for Linear and Nonlinear Relations Identifying causal relations from purely observational data typically requires addi- tional assumptions on relations and/or noise. Most current methods restrict their analysis to datasets that are assumed to have pure linear or nonlinear relations, which is often not reflective of real-world datasets that contain a combination of both. #causaldiscovery #linearrelations #nonlinearrelations

Unlocking the Mystery of Standard Deviations: What is Considered Normal? Learn how 2 standard deviations can define what is considered within the normal range for 95% of the population. Discover how this calculation can determine norms and gain a deeper understanding of statistical significance. #StandardDeviationsExplained #WhatIsNormal #StatisticalSignificance #DataAnalysis101 #Mathematics #StatisticsExplained #NormalDistribution #UnderstandingStandardDeviations #MathNerd #DataScience Full discussion - https://lnkd.in/e5EzZae9

#Day330 of #500DaysGenerativeAI Implementing context enrichment window technique for document retrieval in a vector database which enhances the standard retrieval process by adding surrounding context to each retrieved chunk, improving the coherence and completeness of the returned information.

The first publication in our #specialissue, “Algorithms in Data Classification (2nd Edition)", is now online. Thanks the efforts of the Guest Editor ioannis tsoulos We invite you to read the first paper and submit to this special Issue: https://lnkd.in/gGcd8k5H MDPI University of Ioannina #callforpapers #dataclassification #algorithms

29. Resampling and Imbalanced Classification https://lnkd.in/d-yswDpK

"This is achieved by organizing the graph into multiple layers, with each layer representing a different granularity of the dataset. The top layers contain fewer nodes, which represent clusters of data points, while the bottom layers contain individual data points." #knowledgegraphs #artificialintelligence #searchsystems #HNSWgraphs #HNSW #searchalgorithms #vectordatabases

Founded on the Occam’s razor principle of “the simpler the better,“ a statistical package has been developed by KAUST researchers that optimally adjusts flexible statistical models for spatiotemporal data. The approach, implemented in widely used statistics software, will help researchers to make more accurate predictions from observational datasets. 💽 Visit our dedicated #KAUSTResearch news website via our link-in-bio to read this full story.