Xiangru Shu

More opportunities

12

I still remember the early days of my career as a Machine Learning engineer, navigating various sites looking for data -- Github, Kaggle, UCI, you name it. One particular project comes to mind—a predictive analytics tool for early diagnosis of diabetic retinopathy. Despite the importance and potential impact of the project, we hit a major roadblock: insufficient data. The patient data we had was not only limited but also fraught with privacy concerns. This challenge was an eye-opener, underscoring the critical need for synthetic data in healthcare. Now, that I see multiple services are offering synthetic data. Synthetic data is artificially generated rather than obtained by direct measurement. It mirrors the statistical properties of real-world data, making it a powerful tool in scenarios where data is scarce, sensitive, or difficult to collect. Today we have LLMs, especially GPT-4 that can generate domain-specific (textual) data. Many LLMs are even finetuned and benchmarked using synthetic data generated from GPT-4. But, how can you start and what are some of the best practices? Here are some of the key points that you must consider: 1. Define Clear Objectives: Understand the purpose of generating synthetic data. Whether it's for augmenting training datasets, finetuning like I mentioned preserving privacy, or creating data for underrepresented scenarios, clarity in objectives is crucial. 2. Choose the Right Model: Today, there are various LLMs, according to your budget choose them wisely. GPT-4, Claude, Mistral, LLaMA, etc all good but some are expensive and others are free. If you are low on budget and low on requirements then an open-source model with free Kaggle or Colab support will do. But if you are required to generate a huge amount of data then you might need to spend a little on storage. Keep in mind that running open-source LLMs might require you to spend money because they are very large and require extra storage and computing power. 3. Ensure Data Quality: This is where your model will be tested. Synthetic data should mirror the statistical properties of real-world data. Regular validation against real data ensures the synthetic data’s utility and realism. 4. Maintain Privacy: One of the primary benefits of synthetic data is enhanced privacy. Ensure that the generation process and the synthetic data itself do not inadvertently expose sensitive information. In any case research thoroughly and choose the model according to your requirements. I have attached a code snippet that might help you understand how to generate synthetic and also how to design a productive prompt. The link to the original code can be found in the comment section.

19

7 Comments

I just wanted to share the profiles of a few great folks from the Northwestern University Master's in Data Science program. If anyone in my network is looking to hire some super smart people for your data teams, I highly recommend Lijia Yu, Yukang Wu, and Xinyi Wei!

24

3 Comments

I'm excited to share that I've completed the Big Data with PySpark course! 📝 This course has been an incredible learning journey, deepening my understanding of big data processing, distributed computing, and how to leverage PySpark for efficient data analytics. It has provided me with practical skills that I can immediately apply to real-world projects. A big thanks to everyone who contributed to the course and helped make it a valuable experience! Onward to new challenges and opportunities! #BigData #PySpark #DataAnalytics #LearningJourney #ContinuousLearning

9

60 days ago, I was laid off as a Data Scientist. 60 days ago, I decided to become a solopreneur. 𝐖𝐢𝐧𝐬: - 🖋️ 𝐈 𝐛𝐞𝐜𝐚𝐦𝐞 𝐚 𝐰𝐫𝐢𝐭𝐞𝐫. I hit 125 followers on Medium after publishing 9 articles in 60 days. I set a goal to hit 100 followers by the end of this year. I can't believe I am already there! - 🧠 𝐈 𝐝𝐞𝐯𝐞𝐥𝐨𝐩𝐞𝐝 𝐚 𝐧𝐞𝐰 𝐥𝐞𝐯𝐞𝐥 𝐨𝐟 𝐜𝐮𝐫𝐢𝐨𝐬𝐢𝐭𝐲. I used to hit "fast forward" when an ad came up on a podcast. I used to close the video when "online gurus" started saying "And that's why you should sign up for my course". I used to zone out in a conversation when an uninteresting topic emerged. I used to judge too quickly. With my renewed curiosity, I became a sponge, soaking up every experience: - I actively pay attention to ads on podcasts, online, on the tube and on the streets. I am observing: How do certain brands and people deliver their message? Who do they target? Why did they choose this particular medium? - I become interested in every topic that comes up in a conversation. How does this person tell their story? What can I learn from their journey? What can I do to help? Can I do more research and write about it in an article (with consent)? As it was famously said, “𝑁𝑜𝑡ℎ𝑖𝑛𝑔 𝑏𝑎𝑑 𝑐𝑎𝑛 ℎ𝑎𝑝𝑝𝑒𝑛 𝑡𝑜 𝑎 𝑤𝑟𝑖𝑡𝑒𝑟. 𝐸𝑣𝑒𝑟𝑦𝑡ℎ𝑖𝑛𝑔 𝑖𝑠 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙.” - ❤️ 𝐈 𝐬𝐭𝐨𝐩𝐩𝐞𝐝 𝐛𝐞𝐢𝐧𝐠 𝐬𝐞𝐥𝐟𝐢𝐬𝐡. I overcame the cringe of putting myself out there and creating value in public. I used to think the cringe came from fear. This is partly true. The other part is arrogance. I was arrogant because I believed the world was watching me. It’s not. Nobody knows who I am. Without an audience to serve, I only thought about myself and my precious feelings instead of the impact I can have on other people. I had all these ideas in my head, but I was hoarding them. Why? Because I thought the universe revolved around me and my emotions. Becoming a solopreneur is an act of generosity. 𝐖𝐡𝐚𝐭'𝐬 𝐧𝐞𝐱𝐭? - Writing: I want to reach 1000 followers on Medium and 100 subscribers on Substack by the end of this year. - Consulting: I am building two projects, one on Data Science and the other on Gen AI. I would like to accelerate my building speed and onboard two more projects by the end of September. - Mentoring: I resisted the idea of coaching or mentoring people, but after having a few conversations, I believe I have a lot to offer and guiding others is very enjoyable. The world isn't watching. It's waiting for value. Connect with me: - Substack: https://lnkd.in/eZjhV7Rv - Medium: https://lnkd.in/eYG-jmSv

189

19 Comments

Fully agree to Pascals Summary! (Pascal's summary quote below 👇) The results? Despite the hype around LLMs' impressive performance on medical text processing and licensing exams, they still couldn't beat tried-and-true models like SVM and XGBoost when it came to real-world clinical predictions. This held true regardless of the LLMs' size, prompting strategy, or fine-tuning. So what's going on? The researchers suggest that LLMs may have a "deficiency in clinical reasoning and decision-making" compared to purpose-built models. It's a reminder that, while LLMs are undoubtedly powerful, they're not a magic bullet for every healthcare application. As the study's authors put it: "We call for caution when practitioners adopt LLMs in clinical applications."

1

As a Machine Learning Engineer, I'm always on the lookout for groundbreaking advancements in our field. Time estimation via forecasting has always been a challenging problem in ML, with many layers of complexity and edge cases. I have personally tackled this problem in both the aviation and healthcare industries, and it remains highly relevant (and often unsolved) even as LLMs capture all the attention (hehe, see what I did there). The recent work by the DoorDash ML team on improving ETA predictions by transitioning from tree-based models to multi-task deep learning architectures is truly inspiring. Their well-written and highly relevant article is a must-read for anyone interested in the latest advancements in ML. 🔗 Read the full article #MachineLearning #DeepLearning #AI #TechInnovation #Inspiration

6

📊 Optimizing RAG Applications with TruLens: A Quick Guide Recently explored TruLens for evaluating and fine-tuning Large Language Models (LLMs) in Retrieval Augmented Generation (RAG) applications. Here's what I learned: Key Evaluation Metrics: ✅ Groundedness: Ensures responses are factually supported by context ✅ Answer Relevance: Measures if responses directly address queries ✅ Context Relevance: Evaluates quality of retrieved information The Process: Set up automated evaluation pipelines using TruLens Track performance across different configurations Use metrics to optimize: Chunking strategies Retrieval parameters Context window sizes Embedding models Benefits: Data-driven optimization Automated quality assessment Systematic performance tracking Clear metrics for improvement Result: More accurate, relevant, and reliable RAG applications with measurable improvements in response quality. #AI #MachineLearning #LLM #RAG #DataScience #TruLens https://lnkd.in/gHQnjREi

3

Week 5 CUNY Tech Prep:  I explored more about machine learning and logistic regression this week. I learned how the statistical model can predict between two different outcomes, like pass/fail. While I'll likely use a CNN architecture for my team project, the video taught me some key concepts, such as the confusion matrix and log-odds ratio. I'm thrilled to learn more! https://lnkd.in/e7-ZBf8Q

4

Quentin H. and Yanwei (Wayne) Zhang on the Faire Data team recently did some great work fine-tuning Llama3 model to enhance search relevance. Here are some key insights from the work: • While prompt engineering is valuable, LLM fine-tuning is essential for a domain specific task like understanding semantic search relevance. • Fine-tuned open-source LLMs like Llama3 can deliver impressive performance, scalability, and cost-efficiency compared to using proprietary models. • Effective application of a fine-tuned LLM demands a well-defined problem, robust, high-quality labeled data, and ongoing model refinement. Ground-truth labeled data is crucial for maximizing model performance. • Fine-tuned Llama3-8b achieves performance on par with the larger Llama2–13b model. Read the blog post to dive deeper into these findings! https://lnkd.in/ecP_3E7S We are also hiring (https://lnkd.in/dHfRZcPi) #LLM #AI #Finetuning #LLAMA3 #SearchRelevance #Hiring

93

3 Comments

Data Scientists, This is BIG! Snowflake's Arctic LLM is Here – and Open-Source! 🤯 Get your hands on it now: Interactive Demo: Explore Arctic's capabilities with this Streamlit app → https://lnkd.in/gJ7y38Ng Replicate: Ready to experiment? Find it on Replicate → https://lnkd.in/gyDXiMju Why This Matters: Arctic lets you quickly build demos of your models. This is a game-changer for showcasing your work and making your models more accessible! Huge thanks to Snowflake for making this happen! Are you excited to try Arctic? Let me know in the comments! 👇 #LLM #OpenSource #DataScience #Streamlit #Snowflake

8

📬 In collaboration with Vu Trinh, we present a deep dive into MapReduce. MapReduce may be old and not really needed today, but I would still encourage folks to understand the technology as it's one of the game changing and core data processing tech which inspired a lot of innovation that we see today. 👉https://lnkd.in/evWP2uh8 #dataengineering

95

4 Comments

Microsoft's open-source repo on GraphRAG is gaining significant traction, If you're wondering if it suits your needs, here's an overview. What? GraphRAG uses LLMs to construct knowledge graphs from data and answer user queries based on private datasets. Unlike traditional RAG methods that rely on vector similarity, GraphRAG leverages LLM-generated knowledge graphs to significantly enhance question-and-answer performance, especially for complex document analysis. Why graphs? Graphs excel at connecting different pieces of information through their relationships, making them ideal for synthesizing insights and performing complex analytical tasks, unlike flat, unconnected data. When should you use it? GraphRAG could be ideal for scenarios requiring deep information discovery and analysis across multiple noisy documents or datasets containing mis/ dis-information. GraphRAG can also be beneficial in scenarios where a straightforward semantic search based solely on the query may not suffice. The answer may reside in related information rather than the query term alone. When to skip it? GraphRAG's effectiveness depends on well-structured indexing, particularly in unique datasets with domain-specific concepts. Indexing can be resource-intensive, requiring careful setup and testing before extensive use. GraphRAG's effectiveness also hinges on its ability to accurately identify nodes related to a query, a task that is super straightforward in semantic search. You can always combine both semantic and graph-based methods to leverage the strengths of each approach. link - https://lnkd.in/gFqzRp4i Microsoft #graphrag #microsoft

7

I couldn’t agree more with this post. Data preparation is crucial because even the best algorithms can't compensate for poor-quality data. Thorough data cleaning and preparation are often the biggest factors in the success of a data science project. It’s not just about using advanced models but ensuring the data is reliable and well-structured to get meaningful results.

1

Data is often used to make informed decisions, but understanding causality is the gold standard for gaining deeper insights. In a blog by data scientists at Urban Company, the team shares how they used causal inference methodologies to make more accurate, data-driven decisions. Urban Company is a marketplace that connects consumers with trained professionals for home services. When a customer selects a service, the company may block certain time slots if they anticipate no professional will be available. If a slot isn't blocked but no professional accepts the service request, it's called a 'No Response.' The blog presents case studies where the team estimated the impact of 'No Response' events to determine when they should proactively block slots. Initially, simple aggregation suggested these events led to a 25% drop in customer lifetime value (LTV). However, after applying causal machine learning (ML) methods, the team discovered that the true causal effect was only half that estimate. Confounding factors had exaggerated the initial results, highlighting the importance of using causal ML to address such biases. While the blog doesn't delve deeply into the specific causal ML methodologies like T-learners and X-learners, the appendix offers a solid introduction to these key concepts and formulas. This makes the blog a valuable guide for understanding how to apply causal inference in practice. #datascience #causality #LTV #causalML #machinelearning #datadrivendecision – – –  Check out the "Snacks Weekly on Data Science" podcast and subscribe, where I explain in more detail the concepts discussed in this and future posts:    -- Spotify: https://lnkd.in/gKgaMvbh   -- Apple Podcast: https://lnkd.in/gj6aPBBY    -- Youtube: https://lnkd.in/gcwPeBmR https://lnkd.in/gGqKKstJ

70