A novel approach for the detection of digital facial manipulation

Image for post
Image for post
Photo by Jefferson Santos on Unsplash

The state-of-the-art of image manipulation techniques allows anyone to alter images and videos in order to swap a person’s identity. Techniques like Deepfake and Face2Face are two relatively novel techniques which have gained popularity recently. In addition to these techniques, a variety of GAN-based face-swapping methods have also been published with accompanying code.

There is an ethical factor around the utilization of DeepFakes andFace2Face; As these techniques keep improving the ethical risks associated to them keep growing as well. Consequently, to counter this emerging ethical threat, Facebook Research created the “DeepFake Detection Challenge (DFDC) Dataset” and a Kaggle competition to…

Top 3 tools for a data scientist in academia

Image for post
Image for post
Photo by Campaign Creators on Unsplash

The field of data science moves at a different speed than other areas. Machine learning is constantly evolving and libraries like PyTorch and TensorFlow keep improving. Research companies like Open AI and Deep Mind keep pushing the boundaries of what machine learning can do( i.e.: DALL.E and CLIP). On a basic level, the skills required to be a data scientist remain the same, namely statistics, Python/R programming, databases, PyTorch/TensorFlow and Data Visualization. However, the tools which data scientist use are always changing or being updated.

I work as a research fellow in academia and I have noticed that academia lacks…

Key factors affecting deep learning recommendation models

Image for post
Image for post
Photo by Shahadat Rahman on Unsplash

In a study published by Facebook AI Research (more here) the training efficiency of deep learning models are discussed showcasing data about model architecture and physical infrastructure which is relevant for both industry and academia.

Importance of efficiency of deep learning models

Deep learning models are widely used by tech giants. According to Facebook AI Research, the following companies utilise deep learning recommendation models: Google, Microsoft, Netflix, Alibaba, Facebook/Instagram. Furthermore, companies like Tesla Motors and Waymo Team are doing impressive work with PyTorch and TensorFlow. A glimpse of the work from Tesla can be seen in the following video from Andrej Karpathy on his work on Tesla…

Image for post
Image for post
Photo by NeONBRAND on Unsplash

The State of AI report is produced by AI investors Nathan Benaich and Ian Hogarth. It analyses the most recent and relevant developments in AI and is updated at the end of every year. If you call yourself data scientist or machine learning engineer this report is a must in your reading list.

Full State of AI Report Here

This report covers five thematic areas, namely, Research, Talent, Industry, Politics and Predictions. …

Image for post
Image for post

With each passing day, the number of python notebooks keeps growing. Developers have embraced Jupyter Notebook into their lives and now it has become a must-have in the industry. Every day I keep finding a lot of useful notebooks rather than a regular .py file since many developers are adopting this relatively new paradigm.

I think the first time this called my attention was in 2019, but back then I did not feel inspired enough to implement it. In 2020 I saw many more developers embracing Jupyter Notebooks and a lot of documentation addressing how their code works in a…

A naive money-oriented idea?

Image for post
Image for post
Photo by Kevin Ku on Unsplash

One afternoon, in the middle of my holidays the thought of using machine learning to predict football results in the premier leagues came to my mind. I have never bet on sports myself because I do not like to dispose of the money, I make that way. However, I entertained the idea. I thought that if I can design an algorithm that gives me over 60% accuracy, I could spread the risk and bet on multiple matches, thus making constant revenue.

I thought the following, if I start with £100 (I live in the UK), and bet £10 pounds in…

The AI developed by Google detects breast cancer with higher accuracy

Image for post
Image for post

The goal of an AI is to create algorithms, robots, and technology capable of functioning in an intelligent (human-like) manner. One of Google’s AI tools has shown skills of detection of breast cancer which are similar if not better than those of a trained doctor.

In a study published in Nature (see here), An Artificial Intelligence (AI) developed by Google has improved the early detection process of breast cancer, reducing false negatives and false positives.

The problem

Breast cancer is the second leading cause of death from cancer in women. One of the key aspects of Breast cancer is early detection, which…

A conceptualization of the term smart device

Image for post
Image for post

Smart devices play a fundamental role in today’s Industry 4.0. They are at the center of the Internet of Things (IoT) and smart cities. Smart devices were the center of my PhD investigation a few years ago. However, when I was trying to find a definition for smart devices, not many sources showed up.

I could only find one Wikipedia article with a not certain definition. I resorted to creating a definition for a smart device. I create a methodological and replicable approach for developing a scalable concept of smart device, ending up with the following definition:

A smart…

Key aspects of autonomous vehicles

Image for post
Image for post
Source: Machine Design

When thinking about autonomous vehicles we consider the idea of a car that can take us from origin to destination without the need for human intervention.

Various levels of autonomy have been defined by SAE International (Society of Automotive Engineers). Ranging from level 0 which are our daily-use utility vehicles, until level 5 which are vehicles where no human interaction is required. Currently, we are far from a level 5 automation since there are many situations in which autonomous vehicles are not good dealing with. But if the industry and academia keep the good work we might be there sooner…

Manuel Silverio

Researcher of Connected Autonomous Vehicles (CAVs) at Coventry University, UK. PhD in Digital Transformation.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store