Structured Data

Idea

“If you have videos from people dancing, how can you measure how similar their movements are?”

‍

‍

Context of the internship

‍

• Looking into time-sensitive metrics
• Multivariate time series problem: multiple limbs, multiple x/y anchors in each frame → how to integrate into a single score?
• Technique is more widely applicable:

• similarity search
• clustering
• anomaly detection
• pattern discovery

‍

What you’ll learn

‍

• How to research and implement state of the art time series techniques and similarity metrics while taking into account latency/accuracy requirements.
• How to showcase your work by creating an interactive demo on our website and a blogpost.
• Getting familiar with how a top notch AI consultancy firm operates internally.

‍

Idea

“Where do we as an international company decide to meet-up in order to emit as least CO2 as possible.”

‍

‍

Context of the internship

‍

• Many companies nowadays have the culture of gathering all colleagues at least once a year. These companies are often not taking into account the burden these transportations have on our planet.
• Based on the location of each ML6 agent, we want to automatically be able to determine the best spot to meet up, taking into account rules and constraints.
• The goal is to create a demo that works for an arbitrary number of locations we can showcase on our website as well as a blogpost. 

‍

What you’ll learn

‍

• How to research and implement state of the art optimization techniques while taking into account latency and constraints.
• How to showcase your work by creating an interactive demo on our website and a blogpost.
• Getting familiar with how a top notch AI consultancy firm operates internally.

‍

It’s the contemporary alchemist’s wet dream: predicting financial markets. Impossible or not, the latest edition of the M competitions is all about exactly that: predict the stock prices for a selection of 100 stocks for the next month.

‍

The M competitions are the most influential and widely cited competitions in the field of forecasting. So, of course we’re participating with a team from ML6.

‍

‍

Methodology / Tasks

During this internship, you will:

• Try out a bunch of different time series algorithms and approaches for tackling the problem of stock forecasting
• Approach the problem from the data side: find useful data sources that are related to a single stock
• Experiment with different python packages for time series forecasting
• Try to pick one specific stock and optimize the methods for this specific one
• Follow best practices for creating repeatable experiments by making use of google cloud tools like vertex ai pipelines
• Participate in the M6 competition and hand in your forecasts every month

Profile / Required skills

• Strong analytical abilities, knowledge of different statistical methods, not scared by mathematics and a familiarity with research studies.
• Strong interest in time series algorithms
• Familiarity with statistical analysis languages and tools like Python
• Excellent verbal and written communication in English.
• You are currently pursuing a degree in computer science or related field.

Internship Duration

The duration of the internship can be flexible and depends on the candidate preference and the project requirements. The typical duration is 6 to 8 weeks.

‍

Idea

“Currently, the state-of-the-art similarity metrics are only implemented in R. We want to port these to Python and implement these in frameworks like Tensorflow.”

‍

‍

Context of the internship

‍

• The Wasserstein distance has been around for centuries but recently is causing a furore in ML. In essence, you calculate how different two distributions are, and the result is a number between 0 and +inf.
• Now, we can use the Wasserstein distance as a metric to calculate the degree of difference between two probabilistic functions, but we have to go with a parametric version of it on real life data to estimate the actual Wasserstein distance of the two underlying distributions.
• The question that pops up is: How do we define when 2 distro's are different using the Wasserstein distance? How do we go about hypothesis testing? 🤔

• We are not the first ones to think about this. Schefzik et al. have come up with a way to test this and implemented it in R.

• So... We want to make this test available in python and add it to scipy and TensorFlow Data Validation.

‍

What you’ll learn

‍

• State-of-the-art metrics used extensively in GANs, similarity search, clustering, anomaly detection and pattern discovery;
• How to contribute impactfully to the open-source community and create a blogpost;
• Getting familiar with how a top notch AI consultancy firm operates internally.

‍

The dream for every (process) manager is to have a cockpit that monitors the processes in her supervision and that notifies when an incident happens. Just like when an airplane pilot gets notified when the plane is approaching the ground at high speeds for example.

‍

Lots of big companies with big processes use software to register all the steps that happen in a process. They often use SAP for that.

‍

‍

In this internship, you will

• start with talking to a process manager at a client of ours and discover which processes they run and what they would like to see in their cockpit.
• Then you will dive into their enterprise software (SAP) to extract the relevant data and events.
• Using that data, you will explore machine learning algorithms (unsupervised or supervised) and process mining techniques that are able to detect the relevant metrics and events that will end up in the cockpit dashboard.
• If time (and interest) allows, you will then work on the frontend of the dashboard itself to shape the manager’s experience. The cherry on the cake is that the dashboard allows input from the end user s.t. it functions as a labeling tool s.t. the entire setup can become an active learning system.

‍

Profile / Required skills

• Strong analytical abilities, knowledge of different statistical methods, not scared by mathematics and a familiarity with research studies.
• Familiarity with statistical analysis languages and tools like Python, SQL.
• Excellent verbal and written communication in English.
• You are currently pursuing a degree in computer science or related field.

‍

Internship Duration

The duration of the internship can be flexible and depends on the candidate preference and the project requirements. The typical duration is 6 to 8 weeks. 

‍