Creating a community (LinkedIn group) for my blog - where you can ask me questions re AI

A community (LinkedIn group) for my blog - where you can ask me questions re AI

As you know from this newsletter, I have shared many detailed posts here. 

These relate mostly to my teaching at #universityofoxford and my work in AI. 

My interest in AI ranges from 

1. Teaching at the university of Oxford (AI and AI for engineering sciences) 
2. AI Policy including Responsible AI
3. large scale AI deployments I am involved in (and we are recruiting) and 
4. Maths for AI. 

In 2025, these interests will expand

I always wanted to create a conversation around my blogs but was not sure about the exact mechanism. I tried various things like email newsletters but these are expensive and old fashioned and not a community. I also wanted to create a community around my blog ie a content led approach. Most importantly, I wanted to implement a version of Kevin Kelly’s 1000 true fans i.e. a smaller group of people you actively engage with rather than a vast mailing list

Here is how the community would look like

1. The community  would be an invite only Linkedin Group
2. There is no cost
3. It would comprise of FAQs from my blog (see below)
4. Launched in first week of Jan
5. You can ask questions related to AI and I will try and respond within the week
6. About the FAQ - please see below
7. I got this idea from Google notebookLM. Instead of doing a podcast - its an FAQ  
8. You can apply here https://forms.gle/RQd5Wo4ZLCUowpdbA

Converting blog posts to FAQ

The community would have FAQs of my blog posts

Consider this recent blog on feature engineering from a mathematical point of view

The two central ideas of the blog are 

1. "You can also think of feature engineering as the ability to understand the underlying distribution for the purpose of inference. The underlying distribution of a phenomenon refers to the statistical properties (mean, variance, skewness, relationships, etc.) of the data that generate observations.Inference in machine learning refers to the model’s ability to make predictions or draw conclusions from data. We can think of feature engineering as a bridge between the underlying statistical properties of the data (as represented by its data distribution) and its predictive performance." and 

2. "Feature engineering is, in essence, the process of translating the insights from EDA into features that improve the model’s inference capabilities." on this basis create a FAQ for this blog

Based on this, we get the following FAQ 

PS while I used LLMs to get the FAQ - it does not work on the first shot. You have to prompt it against the key ideas and themes to get a useful FAQ.

FAQ: Understanding Feature Engineering from a Mathematical Perspective

1. What is feature engineering, and why is it important?

Feature engineering is the process of transforming raw data into meaningful features that improve the predictive performance of machine learning models. It acts as a bridge between the underlying statistical properties of data (e.g., mean, variance, skewness) and a model’s inference capabilities.

2. How does feature engineering relate to understanding data distribution?

Feature engineering involves analyzing the underlying distribution of a dataset, including its statistical properties, relationships, and patterns. By understanding this distribution, we can create features that better represent the data and enhance a model's ability to infer or predict outcomes.

3. What is meant by "underlying distribution" in the context of feature engineering?

The underlying distribution refers to the statistical characteristics of the data that generate observations, such as its mean, variance, skewness, and correlations. These properties influence how features are engineered to capture meaningful insights for a model.

4. How does feature engineering act as a bridge between statistical properties and predictive performance?

Feature engineering translates insights from the data’s statistical properties into actionable features that a machine learning model can use. For example, identifying skewness in a distribution might lead to applying log transformations to stabilize variance, thereby improving model performance.

5. What is the role of exploratory data analysis (EDA) in feature engineering?

EDA is a crucial precursor to feature engineering. It helps uncover patterns, anomalies, and relationships within the data. Feature engineering builds on these insights by converting them into features that enhance a model's inference capabilities.

6. Why is inference important in machine learning, and how does feature engineering improve it?

Inference refers to a model’s ability to make predictions or draw conclusions from data. Feature engineering improves inference by ensuring the model has access to features that are well-represented and relevant to the underlying problem, leading to better predictions.

7. Can you provide an example of how EDA insights are translated into features?

Suppose EDA reveals that a variable has a skewed distribution. This insight can guide feature engineering to apply a log transformation, creating a feature that normalizes the distribution and enhances the model’s ability to learn patterns effectively.

8. How does understanding relationships between variables influence feature engineering?

Understanding relationships, such as correlations or interactions between variables, can lead to the creation of composite features. For example, if two variables have a strong interaction effect, feature engineering might involve creating a product or ratio of these variables to capture the relationship.

9. How does feature engineering balance domain knowledge and mathematical techniques?

Feature engineering requires a combination of domain expertise (to identify relevant features and patterns) and mathematical techniques (to apply transformations like scaling, binning, or encoding). Together, these ensure features are meaningful and improve model performance.

Conclusion

You can apply here https://forms.gle/RQd5Wo4ZLCUowpdbA

In the community you can also ask your own questions, in the sense the FAQ will expand in its own right