Ana CALIN

Nmap Cheatsheet

32

The new version of Python is out ! Check the news for f-strings in this caroussel.

13

2 commentaires

Last week I discovered the hard way that large language models are not deterministic: I was trying to optimise a set of prompts for a document classification and couldn't get consistent results, despite trying the full range of model configuration options (temperature, seed, etc...). 𝗛𝗼𝘄 𝗶𝘀 𝘁𝗵𝗮𝘁 𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝗲, 𝘄𝗵𝗲𝗻 𝗮𝗻 𝗟𝗟𝗠 𝗶𝘀 𝗯𝗮𝘀𝗶𝗰𝗮𝗹𝗹𝘆 𝗮 𝗷𝘂𝘀𝘁 𝗮 𝗺𝗮𝘁𝗵 𝗳𝘂𝗻𝗰𝘁𝗶𝗼𝗻 𝘄𝗶𝘁𝗵 𝗮 (𝘃𝗲𝗿𝘆) 𝗹𝗼𝗻𝗴 𝗹𝗶𝘀𝘁 𝗼𝗳 𝗽𝗮𝗿𝗮𝗺𝗲𝘁𝗲𝗿𝘀 ? 𝗔𝘀 𝗶𝘁 𝗶𝘀 𝗼𝗳𝘁𝗲𝗻 𝘁𝗵𝗲 𝗰𝗮𝘀𝗲 𝗶𝗻 𝗔𝗜 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴, 𝗶𝘁 𝗶𝘀 𝗻𝗼𝘁 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝗺𝗮𝘁𝗵, 𝗯𝘂𝘁 𝘁𝗵𝗲 𝗶𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝘁𝗵𝗮𝘁 𝗺𝗮𝗸𝗲𝘀 𝗶𝘁 𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝗲 𝘁𝗼 𝗲𝘅𝗲𝗰𝘂𝘁𝗲 𝗶𝘁. Its core computational units, the GPUs, have thousands of small processors (cores) that work together to perform calculations in parallel.This introduces an inherent level of randomness in the order and timing of operations. As a results LLMs are inherently non deterministic. 𝗟𝗲𝘀𝘀𝗼𝗻 𝗹𝗲𝗮𝗿𝗻𝘁: 𝗱𝗼𝗻'𝘁 𝘄𝗮𝘀𝘁𝗲 𝘁𝗶𝗺𝗲 𝗼𝘃𝗲𝗿-𝗼𝗽𝘁𝗶𝗺𝗶𝘀𝗶𝗻𝗴 𝘆𝗼𝘂𝗿 𝗽𝗿𝗼𝗺𝗽𝘁𝘀 ! #genai #llms #aiengineering Luke Gassmann James Hope

61

9 commentaires

Baal, our Bayesian Active Learning library is working on a major version and we want to know more about you! If you use Baal for Active Learning, Uncertainty Estimation or Bayesian Deep Learning, we would **love** to talk to you! 😎 In more detail, we want to understand when our users use our library and how. You can take a spot in our Calendly:

14

Excited to share the development of an agentic Retrieval-Augmented Generation (RAG) system using LlamaIndex, leveraging both the phi-3-small model and GPT-3.5 Turbo, to predict the outcomes of EURO 2024! ⚽️ The model forecasts France as the likely winner🏆. Group F has been particularly intriguing, with Turkey and Portugal fiercely competing for the top spot. The model suggests that if Portugal wins the group, they advance to the semi-finals, otherwise, they face elimination by France in the Round of 16. The first match to be played today features the host nation Germany against Scotland. The predicted score is Germany vs. Scotland: Germany wins (3-1). It will be exciting to see whether the LLM's prediction is accurate or totally off! #AI #MachineLearning #EURO2024 #SportsAnalytics #LLM #Innovation #LlamaIndex #Microsoft #OpenAI #Betting

26

Predicting Income Levels: A Logistic Regression Analysis in R As many of you might know, Logistic Regression is a statistical method used to model the probability of a binary outcome. In this project, I worked with the UCI Adult dataset, attempting to predict individuals’ classification based on their salary. For an in-depth exploration, please access the link provided below. #DataScience #MachineLearning #DataAnalysis #PredictiveModeling #LogisticRegression #RProgramming #DataVisualization #StatisticalAnalysis #IncomePrediction #BinaryClassification #DataCleaning #ExploratoryDataAnalysis #ModelEvaluation #FeatureEngineering #DataPreprocessing #StatisticalModeling #DataInsights #DataMining #DataAnalytics #ProgrammingTutorial

15

⚡ 🔥 𝐃𝐞𝐞𝐩 𝐆𝐚𝐮𝐬𝐬𝐢𝐚𝐧 𝐏𝐫𝐨𝐜𝐞𝐬𝐬𝐞𝐬: 𝐀 𝐒𝐮𝐫𝐯𝐞𝐲 📘 😎 - a great short paper about #gaussianprocesses ❄ Gaussian processes are one of the dominant approaches in Bayesian learning. Although the approach has been applied to numerous problems with great success, it has a few fundamental limitations. ❄ Multiple methods in literature have addressed these limitations. However, there has not been a comprehensive survey of the topics as of yet. Most existing surveys focus on only one particular variant of Gaussian processes and their derivatives. ❄ This survey details the core motivations for using Gaussian processes, their mathematical formulations, limitations, and research themes that have flourished over the years to address said limitations. ❄ Furthermore, one particular research area is Deep Gaussian Processes (DGPs), it has improved substantially in the past decade. The significant publications that advanced the forefront of this research area are outlined in their survey. ❄ Finally, a brief discussion on open problems and research directions for future work is presented at the end. #deeplearning #bayesian

96

1 commentaire

Day 7 of Deep learning 𝐁𝐚𝐜𝐤𝐩𝐫𝐨𝐩𝐚𝐠𝐚𝐭𝐢𝐨𝐧 𝐢𝐧 𝐃𝐞𝐞𝐩 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠! 🧠💡 🔍 𝐖𝐡𝐚𝐭 𝐢𝐬 𝐁𝐚𝐜𝐤𝐩𝐫𝐨𝐩𝐚𝐠𝐚𝐭𝐢𝐨𝐧? Backpropagation is the backbone of training artificial neural networks, including deep learning models. It's a crucial algorithm that enables networks to learn from data, adapt, and improve over time. 🔄 𝐇𝐨𝐰 𝐝𝐨𝐞𝐬 𝐢𝐭 𝐰𝐨𝐫𝐤? 1️⃣ 𝑭𝒐𝒓𝒘𝒂𝒓𝒅 𝑷𝒂𝒔𝒔: Data is fed into the network, propagating through layers, producing an output. 2️⃣ 𝑳𝒐𝒔𝒔 𝑪𝒂𝒍𝒄𝒖𝒍𝒂𝒕𝒊𝒐𝒏: The output is compared to the actual target, and a loss function quantifies the performance. 3️⃣ 𝑩𝒂𝒄𝒌𝒘𝒂𝒓𝒅 𝑷𝒂𝒔𝒔 (𝑩𝒂𝒄𝒌𝒑𝒓𝒐𝒑𝒂𝒈𝒂𝒕𝒊𝒐𝒏): Gradients of the loss function with respect to network parameters (weights and biases) are computed using the chain rule of calculus. 4️⃣ 𝑮𝒓𝒂𝒅𝒊𝒆𝒏𝒕 𝑫𝒆𝒔𝒄𝒆𝒏𝒕: Weights are updated iteratively using optimization algorithms like gradient descent, minimizing the loss function. 💡 𝐖𝐡𝐲 𝐢𝐬 𝐢𝐭 𝐢𝐦𝐩𝐨𝐫𝐭𝐚𝐧𝐭? Backpropagation allows networks to learn from mistakes, adjusting parameters to improve performance on tasks like image recognition, natural language processing, and more. 🔬 𝐂𝐨𝐧𝐭𝐢𝐧𝐮𝐨𝐮𝐬 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠: Understanding backpropagation unlocks the potential for building more accurate and efficient deep learning models, driving innovation in various domains. #DeepLearning #ArtificialIntelligence #NeuralNetworks #Backpropagation #AIInnovation

2

Rentrez dans les arcanes de #python en explorant les secrets de la library Pandas : "Spoiler alert, it is mostly about operator overloading and predicates!" https://hubs.ly/Q02Cpy2t0 #Data #DataAnalysis #SoftwareDevelopment #AdvancedProgramming #TechSkills #BehindTheScene

2

1 commentaire

🔍 𝐇𝐨𝐰 𝐢𝐬 𝐂𝐥𝐚𝐮𝐝𝐞 3.5 𝐒𝐨𝐧𝐧𝐞𝐭 𝐓𝐫𝐚𝐧𝐬𝐟𝐨𝐫𝐦𝐢𝐧𝐠 𝐇𝐚𝐧𝐝𝐰𝐫𝐢𝐭𝐢𝐧𝐠 𝐀𝐧𝐚𝐥𝐲𝐬𝐢𝐬? In the rapidly evolving field of artificial intelligence, Claude 3.5 Sonnet is setting new benchmarks in vision and analysis capabilities. Here's how this advanced AI model is revolutionizing handwriting analysis: 📝 𝐄𝐧𝐡𝐚𝐧𝐜𝐞𝐝 𝐕𝐢𝐬𝐢𝐨𝐧 𝐂𝐚𝐩𝐚𝐛𝐢𝐥𝐢𝐭𝐢𝐞𝐬: Claude 3.5 Sonnet, with its cutting-edge vision capabilities, is enabling users to upload images of handwriting and receive detailed personality interpretations, showcasing a significant leap in visual data processing. 🚀 𝐒𝐩𝐞𝐞𝐝 𝐚𝐧𝐝 𝐄𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐜𝐲: In head-to-head tests with GPT-4, Claude 3.5 Sonnet has demonstrated superior speed, especially in tasks requiring high-level reasoning and multilingual capabilities, making it a preferred choice for developers and researchers. 💡 𝐂𝐨𝐬𝐭-𝐄𝐟𝐟𝐞𝐜𝐭𝐢𝐯𝐞𝐧𝐞𝐬𝐬: Despite its advanced features, Claude 3.5 Sonnet is surprisingly affordable. It's available for free on Claude's platform, with competitive pricing for high-volume usage, making it accessible to a broader audience. 🛠️ 𝐔𝐬𝐞𝐫-𝐅𝐫𝐢𝐞𝐧𝐝𝐥𝐲 𝐈𝐧𝐭𝐞𝐠𝐫𝐚𝐭𝐢𝐨𝐧: With an API that echoes the simplicity of the OpenAI API, transitioning to using Anthropic's technology is nearly seamless for developers familiar with existing AI models. 🖥️ 𝐑𝐞𝐚𝐥-𝐭𝐢𝐦𝐞 𝐂𝐨𝐥𝐥𝐚𝐛𝐨𝐫𝐚𝐭𝐢𝐨𝐧: The introduction of 'artifacts' in Claude allows users to interact with AI-generated content like code snippets and text documents directly, enhancing the collaborative experience and integrating AI insights more fluidly into projects. The application of Claude 3.5 Sonnet in handwriting analysis not only demonstrates the model’s robust capabilities but also hints at the potential future applications in various industries requiring nuanced visual interpretations. 🤔 Could Claude 3.5 Sonnet be the key to unlocking new dimensions in your field of work? How might enhanced AI vision capabilities impact your industry? #AI #MachineLearning #HandwritingAnalysis #Innovation #Claude35Sonnet

2

🚀 Let’s understand the Neural Network Learning Process! 🚀 In this diagram, we delve into how a neural network learns to recognize different animal species, from basic shapes to complex features. 1. Training: Initially, the neural network is trained on a large dataset of labeled images, where each image is associated with the corresponding animal. 2. Input: An unlabeled image is fed into the network. 3. First Layer: Neurons in the first layer detect basic shapes and edges within the image. 4. Higher Layers: As the image progresses through the network, neurons in higher layers respond to increasingly complex features, such as eyes, ears, and fur patterns. 5. Top Layer: Finally, the top layer neurons fire based on the most intricate features, allowing the network to identify the animal in the image with high accuracy. 6. Output: The network produces a prediction of the animal’s identity, along with a confidence score. 🌟 Key Takeaways: ◾Neural Networks Learn From Data: The more data a network is trained on, the better it becomes at recognizing patterns and making accurate predictions. ◾Hierarchy of Layers: Different layers process information at varying levels of abstraction. ◾Adaptability of Neural Networks: They can learn new things and improve performance over time. 🚀 This is a growing field with a lot of potential to revolutionize our way of life and work. 🚀 🤖Feel free to share your thoughts on neural network learning!🤖 For learning materials & resources in Data/ML/AI, please check my previous posts.   Please make sure you’ve clicked on the ‘follow’ button and let’s learn together! Naguib MAOULIDA   #technology #it #artificialintelligence #machinelearning #deeplearning #data #datascience #dataintelligence #dataanalysis #dataengineering #datapipeline #dataintegration #datagovernance #datastrategy #etl #elt #reverseetl #bigdata #generativeai #ai

6

I am currently taking a fundamental #statistics course and it just makes me appreciate all the methods in #machinelearning even more. Here are a summary of the topics you'd want to look into in a beginners course on statistics : 1. Moment-generating functions 2. Joint MGFs and Delta functions 3. Bivariate Delta methods and Continuity-Theorem 4. Central Limit Theorem 5. Chi-Squared, T- and F-distributions 6. Estimator bias 7. Methods of moments estimators, Maximum Likelihood Estimation 8. Cramer-Rao inequality, Fischer Information 9. Order Statistics, Confidence Intervals 10. Hypothesis testing and Neyman-Pearson lemma #datascience #mathematics

62

9 commentaires

GraphRAG : New tool for complex data discovery GraphRAG is a structured, hierarchical approach to Retrieval Augmented Generation (RAG), as opposed to naive semantic-search approaches using plain text snippets. The GraphRAG process involves extracting a knowledge graph out of raw text, building a community hierarchy, generating summaries for these communities, and then leveraging these structures when perform RAG-based tasks. #LLM #GenAI #Graph #RAG #graphrag #AI Reference : https://lnkd.in/gYbw5yXG

5

𝟮𝟱 𝗠𝗼𝘀𝘁 𝗜𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝘁 𝗠𝗮𝘁𝗵𝗲𝗺𝗮𝘁𝗶𝗰𝗮𝗹 𝗗𝗲𝗳𝗶𝗻𝗶𝘁𝗶𝗼𝗻𝘀 𝗶𝗻 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 ⤵️ [𝟭] 𝗚𝗿𝗮𝗱𝗶𝗲𝗻𝘁 𝗗𝗲𝘀𝗰𝗲𝗻𝘁: Minimizes a function by iteratively moving in the direction of steepest descent. [𝟮] 𝗡𝗼𝗿𝗺𝗮𝗹 𝗗𝗶𝘀𝘁𝗿𝗶𝗯𝘂𝘁𝗶𝗼𝗻: Bell-shaped probability distribution. [𝟯] 𝗭-𝘀𝗰𝗼𝗿𝗲: Standard deviations from the mean. [𝟰] 𝗦𝗶𝗴𝗺𝗼𝗶𝗱 𝗙𝘂𝗻𝗰𝘁𝗶𝗼𝗻: Maps any real number to a value between 0 & 1 [𝟱] 𝗖𝗼𝗿𝗿𝗲𝗹𝗮𝘁𝗶𝗼𝗻: Linear relationship between two variables. [𝟲] 𝗖𝗼𝘀𝗶𝗻𝗲 𝗦𝗶𝗺𝗶𝗹𝗮𝗿𝗶𝘁𝘆: Similarity between vectors. [𝟳] 𝗡𝗮𝗶𝘃𝗲 𝗕𝗮𝘆𝗲𝘀: Classifies using Bayes' theorem. [𝟴] 𝗠𝗮𝘅𝗶𝗺𝘂𝗺 𝗟𝗶𝗸𝗲𝗹𝗶𝗵𝗼𝗼𝗱 𝗘𝘀𝘁𝗶𝗺𝗮𝘁𝗶𝗼𝗻: Estimates model parameters for maximum data likelihood. [𝟵] 𝗢𝗿𝗱𝗶𝗻𝗮𝗿𝘆 𝗟𝗲𝗮𝘀𝘁 𝗦𝗾𝘂𝗮𝗿𝗲𝘀: Fits linear regression models by minimizing squared errors. [𝟭𝟬] 𝗙𝟭 𝗦𝗰𝗼𝗿𝗲: Harmonic mean of precision and recall for classification models. [𝟭𝟭] 𝗥𝗲𝗟𝗨 (𝗥𝗲𝗰𝘁𝗶𝗳𝗶𝗲𝗱 𝗟𝗶𝗻𝗲𝗮𝗿 𝗨𝗻𝗶𝘁): Activation function for neural networks. [𝟭𝟮] 𝗦𝗼𝗳𝘁𝗺𝗮𝘅 𝗙𝘂𝗻𝗰𝘁𝗶𝗼𝗻: Outputs a probability vector that sums to 1. [𝟭𝟯] 𝗥-𝘀𝗾𝘂𝗮𝗿𝗲𝗱: Proportion of variance explained by the regression model. [𝟭𝟰] 𝗠𝗲𝗮𝗻 𝗦𝗾𝘂𝗮𝗿𝗲𝗱 𝗘𝗿𝗿𝗼𝗿: Average squared difference between estimated and actual values. [𝟭𝟱] 𝗠𝗲𝗮𝗻 𝗦𝗾𝘂𝗮𝗿𝗲𝗱 𝗘𝗿𝗿𝗼𝗿 𝘄𝗶𝘁𝗵 𝗟𝟮 𝗥𝗲𝗴𝘂𝗹𝗮𝗿𝗶𝘇𝗮𝘁𝗶𝗼𝗻: MSE with penalty for overfitting. [𝟭𝟲] 𝗘𝗶𝗴𝗲𝗻𝘃𝗲𝗰𝘁𝗼𝗿𝘀 & 𝗘𝗶𝗴𝗲𝗻𝘃𝗮𝗹𝘂𝗲𝘀: Eigenvector transforms under a linear transformation by a scaling factor. [𝟭𝟳] 𝗘𝗻𝘁𝗿𝗼𝗽𝘆: Measure of uncertainty in a random variable. [𝟭𝟴] 𝗞-𝗺𝗲𝗮𝗻𝘀 𝗖𝗹𝘂𝘀𝘁𝗲𝗿𝗶𝗻𝗴: Unsupervised learning that groups data points into k clusters. [𝟭𝟵] 𝗞𝘂𝗹𝗹𝗯𝗮𝗰𝗸–𝗟𝗲𝗶𝗯𝗹𝗲𝗿 𝗗𝗶𝘃𝗲𝗿𝗴𝗲𝗻𝗰𝗲: Measures the difference between two probability distributions. [𝟮𝟬] 𝗟𝗼𝗴 𝗟𝗼𝘀𝘀: Loss function for classification problems with probability outputs. [𝟮𝟭] 𝗦𝘂𝗽𝗽𝗼𝗿𝘁 𝗩𝗲𝗰𝘁𝗼𝗿 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 (𝗦𝗩𝗠): Finds a hyperplane with maximum margin for classification. [𝟮𝟮] 𝗟𝗶𝗻𝗲𝗮𝗿 𝗥𝗲𝗴𝗿𝗲𝘀𝘀𝗶𝗼𝗻: Models the relationship between a dependent and independent variable(s). [𝟮𝟯] 𝗦𝗶𝗻𝗴𝘂𝗹𝗮𝗿 𝗩𝗮𝗹𝘂𝗲 𝗗𝗲𝗰𝗼𝗺𝗽𝗼𝘀𝗶𝘁𝗶𝗼𝗻 (𝗦𝗩𝗗): Decomposes a matrix into three component matrices. [𝟮𝟰] 𝗟𝗮𝗴𝗿𝗮𝗻𝗴𝗲 𝗠𝘂𝗹𝘁𝗶𝗽𝗹𝗶𝗲𝗿𝘀: Solves optimization problems with equality constraints. [𝟮𝟱] 𝗕𝗮𝘆𝗲𝘀 𝘁𝗵𝗲𝗼𝗿𝗲𝗺: Relates the conditional and marginal probabilities of random variables. --- Like this post? Follow Muhammad Ehsan, press “like,” and hit the 🔔 on my profile and/or share with your network. #mathematics #datascience #machinelearning #statistics #algorithms #neuralnetworks #deeplearning #probability #linearregression #clustering #classification #optimization #informationtheory #programming #ai #dimensionalityreduction #patternrecognition #mathematicalmodeling #artificialintelligence

2

I do have a question for you though: In your experience, How would you implement a custom autograd function in PyTorch for a new operation, and how would you ensure that both the forward and backward passes are correctly defined?

3

1 commentaire

Gentle introduction of Bayesian Linear Regression by Yuki Shizuya #mathematics #statistics #machinelearning #pymc #python

1