Victor Kostyuk’s Post

Thinking of using reinforcement learning ML to personalize? You're going to need to solve the attribution problem. Our CTO and co-founder Victor Kostyuk explains some of the complications of this thorny problem in our latest post. 👇

Thinking of using reinforcement learning ML to personalize? You're going to need to solve the attribution problem. Our CTO and co-founder Victor Kostyuk explains some of the complications of this thorny problem in our latest post. 👇

Hey Connections 👋 I am sharing another project on Advertisement roles to predict the sales of a company. ✨ Description: This dataset provides information about different advertising scenario such as TV, Newspaper and Radio to predict the sales. ✨ Aim: To build a Multiple Linear Regression model and find the performance parameters. ✨ Steps Implemented: 1. Data Uploading 2. Finding missing values 3. Separation of input and output variables 4. Encoding 5. Normalization 6. Model Creation 7. Performance Evaluation 8. Inference ✨ Inference: A Multiple Linear Regression model is implemented for advertisement and performance parameters such as Mean absolute percentage error and R2 score is calculated. Guide: Sabir K Source: Kaggle IDE: Google Collab Thanking Sabir Sir and Luminar team for the support and guidance.

We've been busy refining our product-market fit with initial customers, including a promising 50-60 person company that could become a key lighthouse client. Product updates: Crawling multiple customer-specified websites Extracting unique insights using LLMs Organizing facts by categories (company overview, pricing, features, use cases) Developing two key outputs: Activity feed for real-time updates Weekly/biweekly newsletter summarizing changes across tracked companies We're focusing on vertical markets, particularly in the observability space, to optimize our resource-intensive process of website crawling and LLM processing. Looking ahead, we're exploring a graph RAG (Retrieval-Augmented Generation) approach. This involves entity and fact extraction to create a traversable graph format, potentially improving accuracy and efficiency in our information processing pipeline.

This map condenses my argument from "Unbundling User Research" into one picture. One that can be re-drawn, augmented, or annotate directly, instead of needing to argue with 2000 words of text. I've omitted quite a bit to show the crux of things, but if you're vaguely familiar with Wardley Maps, I think it gets the point across. Can you grok it? I've explained the annotations with more detail in the latest Loops and Cycles newsletter — link in comments if you'd like to follow this further.

📍 Would you like to know my proven SPIKE Strategy behind this? Comment " SPIKE Strategy" to get PDF free of cost! That Includes: → Secret Niche behind this Graph → Secret to Meet Intent Strategy → CTR Improvement Secret → Algorithms Understanding Behind the CTR Yes it's possible!!!! With proven experience, it's sure that it will work perfectly fine for you.

Machine Learning 😊 github link: https://lnkd.in/gXjMVMhk K-Nearest Neighbour is one of the simplest Machine Learning algorithms based on Supervised Learning technique. K-NN algorithm assumes the similarity between the new case/data and available cases and put the new case into the category that is most similar to the available categories.

We live in a world where algorithms quietly shape our lives in ways we might not even recognize. In this week's newsletter, we explore five examples of how individuals are asserting their agency over algorithms or opening themselves up to algorithm-created possibilities and connections. Algorithms might be powerful, but that doesn’t mean users are powerless. Read more: https://bit.ly/3BNGHWb

𝐋𝐨𝐨𝐤𝐢𝐧𝐠 𝐭𝐨 𝐝𝐢𝐯𝐞 𝐢𝐧𝐭𝐨 𝐬𝐨𝐦𝐞 𝐒𝐮𝐧𝐝𝐚𝐲 𝐫𝐞𝐚𝐝𝐢𝐧𝐠? 𝐇𝐞𝐫𝐞 𝐚𝐫𝐞 𝐟𝐢𝐯𝐞 𝐚𝐫𝐭𝐢𝐜𝐥𝐞𝐬 𝐚𝐥𝐥 𝐚𝐛𝐨𝐮𝐭 𝐜𝐫𝐚𝐟𝐭𝐢𝐧𝐠 𝐲𝐨𝐮𝐫 𝐯𝐞𝐫𝐲 𝐨𝐰𝐧 𝐑𝐀𝐆 (𝐑𝐞𝐭𝐫𝐢𝐞𝐯𝐚𝐥 𝐀𝐮𝐠𝐦𝐞𝐧𝐭𝐞𝐝 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐨𝐧) 𝐬𝐲𝐬𝐭𝐞𝐦: 1. Retrieval Augmented Generation: Best Practices and Use Cases  - Delve into the world of RAG with this comprehensive piece, covering everything from its core components to real-world applications that'll leave you inspired. (https://lnkd.in/di5YVpUJ) 2. Building Performant RAG Applications for Production  - Get ready to optimize your RAG pipelines like a pro! Learn the secrets of fine-tuning for production environments and discover techniques for maximum efficiency.  (https://lnkd.in/dJpWfQsC) 3. Building High-Quality RAG Systems  - Ready to take your RAG game to the next level? This article walks you through the process, highlighting key factors like data loaders and multimedia content handling for top-notch results. (https://lnkd.in/d4BTmzAT) 4. 12 Challenges in Building RAG-based LLM Applications - Buckle up for a journey through the hurdles of RAG-based applications. From reranking retrieval results to handling massive datasets, this article covers it all.  (https://lnkd.in/d2MSz4Yy) 5. 10 Ways to Improve the Performance of Retrieval Augmented Generation Systems  - Ready to supercharge your RAG system? Discover ten actionable strategies to boost performance, from fine-tuning models to harnessing hybrid search techniques.  (https://lnkd.in/dFshxzjb) Grab your favorite beverage, settle in, and prepare to be captivated by the world of RAG!

Ready to unleash the magic of image recognition? 🌟 Dive into our latest project where we explore a captivating dataset featuring our furry friends - cats and dogs! 🐱🐶 With 10,000 train images and a corresponding CSV file, we're embarking on an adventure with 5 unique featuring models and a showdown between 3 powerhouse machine learning algorithms: Random Forest, Naive Bayes, and Gradient Boosting. Get ready to witness the art of predicting pet presence in images come to life! 📸💫 #ImageRecognition #MachineLearningMagic #PetPredictions https://lnkd.in/eE5hzhH8