Jul 31, 2018 · newsletter

Neural reinterpretations of movie trailers

In his latest project, artist and coder Mario Klingemann uses a neural network to match archival movie footage with the content of recent movie trailers. He regularly posts the resulting “neural reinterpretations” on his Twitter. The results are technically impressive. They’re also a fascinating view into how to explore the creative possibilities of a machine learning technique.

Looking through Klingemann’s tweets you can trace his explorations:

![A screenshot from Klingemann’s video of similar scene classification. A 3x3 grid shows several similar looking scenes. Some have planes, others are mostly blank, some have spare drawings of squares.]({{ site.github.url}}/images/editor_uploads/2018-06-26-144731-Screen_Shot_2018_06_25_at_10_45_15_AM.png)

Mario Klingemann’s neural scene classifier grouping scenes it finds similar.

![On the left is a shot of Brad Pitt from Fight Club; on the right is a man holding a telephone with a similar expression from the archive footage.]({{ site.github.url}}/images/editor_uploads/2018-06-26-144942-Screen_Shot_2018_06_25_at_10_46_42_AM.png)

A neural reinterpretation of the Fight Club trailer, with the original footage on the left and the matched on the right.

The movie trailer reinterpretations are a great showcase for the technique for a couple of reasons:

  1. Trailers are made up of short clips. This gives the algorithm lots of shots at finding interesting matches (every cut is a new example). If it was instead focused on a 2 minute long continuous scene, you wouldn’t get to see nearly as many matches. Also the fact that the cuts are often timed to the music makes the reinterpreted content appear more connected to the audio of the trailer.

  2. Films have a built up vocabulary of what different shots mean, like a close-up of a face to signal intense feelings. Film-makers employ these patterns consciously. As film watchers, we may not think about scene types explicitly, but we do build up associations and expectations with different framing, movements, and styles. The side-by-side reinterpretations make this referential language more visible by showing us two examples at a time, helping us notice the similarity the machine has identified. We can then often extrapolate even further into “ah, right, that’s another one of those ‘vehicles rushing by’ shots” that you normally don’t consciously note. This takes the trailers from technical demos into artistic territory.

![A screenshot of the video by Memo Atken. On the left is a blanket being scrunched up by hands; on the right is an image that looks like a painting of waves, where the shape of the waves matches the position of the hands and blanket.]({{ site.github.url}}/images/editor_uploads/2018-06-26-145050-Screen_Shot_2018_06_25_at_10_47_09_AM.png)

A still from “Learning to see: Gloomy Sunday” by Memo Atken

“Learning to see: Gloomy Sunday” by Memo Akten explores similarity in a different, fascinating way. He has a model trained on specific types of art that interpret his webcam photos and generate new images: for example, a sheet becomes waves. Like in the trailer reinterpretations, what takes this beyond technical demo is how suggestive the association can be. The machine’s ability to identify similarity between a sheet and a wave gives us an understanding that we can then apply outside of the context of the video. It’s a suggestive analogy that opens out so that the viewer can build upon it and make their own connections.

Read more

Jul 31, 2018 · newsletter
Jul 24, 2018 · post

Latest posts

Nov 15, 2020 · post

Representation Learning 101 for Software Engineers

by Victor Dibia · Figure 1: Overview of representation learning methods. TLDR; Good representations of data (e.g., text, images) are critical for solving many tasks (e.g., search or recommendations). Deep representation learning yields state of the art results when used to create these representations. In this article, we review methods for representation learning and walk through an example using pretrained models. Introduction Deep Neural Networks (DNNs) have become a particularly useful tool in building intelligent systems that simplify cognitive tasks for users. more
Jun 22, 2020 · post

How to Explain HuggingFace BERT for Question Answering NLP Models with TF 2.0

by Victor · Given a question and a passage, the task of Question Answering (QA) focuses on identifying the exact span within the passage that answers the question. Figure 1: In this sample, a BERTbase model gets the answer correct (Achaemenid Persia). Model gradients show that the token “subordinate ..” is impactful in the selection of an answer to the question “Macedonia was under the rule of which country?". This makes sense .. good for BERTbase. more
Jun 16, 2020 · notebook

Evaluating QA: Metrics, Predictions, and the Null Response →

by Melanie · A deep dive into computing QA predictions and when to tell BERT to zip it! In our last post, Building a QA System with BERT on Wikipedia, we used the HuggingFace framework to train BERT on the SQuAD2.0 dataset and built a simple QA system on top of the Wikipedia search engine. This time, we’ll look at how to assess the quality of a BERT-like model for Question Answering.
May 19, 2020 · notebook

Building a QA System with BERT on Wikipedia →

by Melanie · So you’ve decided to build a QA system. You want to start with something simple and general so you plan to make it open domain using Wikipedia as a corpus for answering questions. You want to use the best NLP that your compute resources allow (you’re lucky enough to have access to a GPU) so you’re going to focus on the big, flashy Transformer models that are all the rage these days.
Apr 28, 2020 · notebook

Intro to Automated Question Answering →

by Melanie · Welcome to the first edition of the Cloudera Fast Forward blog on Natural Language Processing for Question Answering! Throughout this series, we’ll build a Question Answering (QA) system with off-the-shelf algorithms and libraries and blog about our process and what we find along the way. We hope to wind up with a beginning-to-end documentary that provides:
Apr 1, 2020 · newsletter

Enterprise Grade ML

by Shioulin · At Cloudera Fast Forward, one of the mechanisms we use to tightly couple machine learning research with application is through application development projects for both internal and external clients. The problems we tackle in these projects are wide ranging and cut across various industries; the end goal is a production system that translates data into business impact. What is Enterprise Grade Machine Learning? Enterprise grade ML, a term mentioned in a paper put forth by Microsoft, refers to ML applications where there is a high level of scrutiny for data handling, model fairness, user privacy, and debuggability. more

Popular posts

Oct 30, 2019 · newsletter
Exciting Applications of Graph Neural Networks
Nov 14, 2018 · post
Federated learning: distributed machine learning with data locality and privacy
Apr 10, 2018 · post
PyTorch for Recommenders 101
Oct 4, 2017 · post
First Look: Using Three.js for 2D Data Visualization
Aug 22, 2016 · whitepaper
Under the Hood of the Variational Autoencoder (in Prose and Code)
Feb 24, 2016 · post
"Hello world" in Keras (or, Scikit-learn versus Keras)


In-depth guides to specific machine learning capabilities


Machine learning prototypes and interactive notebooks


A usable library for question answering on large datasets.

Explain BERT for Question Answering Models

Tensorflow 2.0 notebook to explain and visualize a HuggingFace BERT for Question Answering model.

NLP for Question Answering

Ongoing posts and code documenting the process of building a question answering model.

Interpretability Revisited: SHAP and LIME

Explore how to use LIME and SHAP for interpretability.


Cloudera Fast Forward is an applied machine learning reseach group.
Cloudera   Blog   Twitter