← Back to all posts

Reading time - 4 mins

1. ML Picks of the Week

A weekly dose of ML tools, concepts & interview prep — all in under 1 minute

2. Technical ML Section

Understand how Bayesian Optimization tune ML models and why to use it.

3. Career ML Section

Learn 3 reasons why growing LinkedIn network is a must for ML career.

1. ML Picks of the Week

🥇ML Tool

Python Library STUMPY

STUMPY allows you to get to your time series insights faster.

Apart from many application, STUMPY works especially well for:

1. repeated pattern identification 2. anomaly detection

Whenever you face these ML problems, try STUMPY first.

📈 ML Concept

Batch Normalization

Batch normalization is a method used in training deep neural networks to:

- stabilize and speed up learning - improve generalization (acts as regularization)

It does this by normalizing the activations (outputs) of a layer for each mini-batch, so they have a consistent distribution during training.

It’s widely used in modern Deep Learning architectures like ResNet & VGG and important concept to know for daily practice and interviews.

Learn more about batch norm HERE & HERE.

🤔 ML Interview Question

What is bias-variance trade-off?

Short answer: The bias-variance tradeoff deals with the balance between:

Model's ability to accurately represent the underlying data patterns (bias)
Model's sensitivity to fluctuations in the training data (variance).

If you train long -> bias is low, variance is high

If you don't train much -> bias is high, variance is low

Learn more: Great B-V Trade Off Article

2. Technical ML Section

1️⃣ What is Bayesian Optimization?

Hyperparameter tuning can make a 30-40% difference in model accuracy.

Why does this happen? This is because, in the search space there are often many local minima, and the difference between them can be quite big.

If you randomly choose a set of hyperparameters or perform a Grid Search, there is a very high probability that you have not investigated other local minima.

Bayesian Optimization solves this by learning from past trials and efficiently selecting the best hyperparameters—saving both time and compute.

2️⃣ How does it work?

Here’s an example of Bayesian Optimization minimizing a cost function.

These are the main steps of Bayesian Optimization for ML model tuning:

📌 Step 1: Set the Objective function, e.g.error on the validation set.

📌 Step 2: Perform the initial sampling of several points of this function.

📌 Step 3: Using these samples, we fit a probabilistic model, e.g. Gaussian Processes (see the gif ↑↑↑).

📌 Step 4. Define an Acquisition Function.

The probabilistic predictions are then used in an acquisition function. In the example above, we use the Lower Confidence Bound (LCB). The next hyperparameter value chosen minimizes the LCB value (see the gif ↑↑↑).

📌 Step 5. Specify the number of iterations to sample.

📌 Step 6. Select the lowest found error and associated hyperparameters

3️⃣ Why to use Bayesian Optimization? Comparison with Grid & Random Search.

Here’s a comparative example. You see that:

Grid Search is exhaustive but inefficient
Random Search ignores past results
Bayesian Optimization adapts and converges faster

2. ML Career Section

3 Reasons to grow your LinkedIn network and optimize profile.

✅ Reason 1: Increased Visibility in LinkedIn Search

Recruiters search for candidates on LinkedIn, but poorly optimized profiles never appear at the top of those results. If you have a strong profile and a large network, your chances of being contacted by recruiters increase significantly.

✅ Reason 2: You Build Trust

Remember: "People recommend or hire those they trust."

Engaging with your network is key. Try sending at least 5 direct messages (DMs) per week to people in your field. Over time, these conversations help build stronger connections and trust, making it much easier to find new opportunities.

✅ Reason 3: You Build a Personal Brand

Personal branding isn’t just for influencers or Hollywood stars. Every professional needs a strong brand to stay competitive.

By growing your LinkedIn network, actively engaging, and posting just once per week, you can already position yourself in the top 5% of professionals on the platform.

That is it for this week!

If you haven’t yet, follow me on LinkedIn where I share Technical and Career ML content every day!

Whenever you're ready, there are 3 ways I can help you:

Join Maistermind for 1 weekly piece with 2 ML guides:

1. Technical ML tutorial or skill learning guide
2. Tips list to grow ML career, LinkedIn, income

Join here!

#7: Why use Bayesian Optimization for ML Model Tuning | 3 Reasons to grow your LinkedIn network

Reading time - 4 mins