nairoz

And don't fly

Really liked it. I'm curious about games for which the guesses are not always a valid solution but can contain special operators. For instance, add a state which is true for 2 different values in Secret Code : 1or2. The code won't contain it but it can be useful for getting more information at each step.

It's really cool. I have always wondered if it would be possible to have video encoders designed for some specific games with prior knowledge about important regions to encode with more details. Example would be the center of the screen for the main character.

Digital Elevation Model (https://en.wikipedia.org/wiki/Digital_elevation_model)

Location: Lyon, France

Remote: Yes

Willing to relocate: Yes (Europe)

Technologies: Python, C++, CUDA, Pytorch

Résumé/CV: https://ncherel.github.io/resume.pdf

Email: nicolas.cherel@outlook.com

I am a recent PhD graduate in machine learning for image and video processing (diffusion models, GANs). I am a research engineer who writes high-quality code, implements state-of-the-art methods, and pushes the boundaries of existing knowledge when necessary. I have worked in large companies, start-ups, and research institutes.

I guess yes. Having worked on video processing, it's always better if you can stabilize because it significantly reduces the number of unique tokens, which would be even more useful for the present method. However, you probably lose in generalization performance and not all videos can be stabilized.

Location: Zurich, Switzerland

Remote: No

Willing to relocate: Yes (Europe)

Technologies: Python, PyTorch, Cuda, C++

Résumé/CV: https://ncherel.github.io/resume.pdf

Email: nicolas.cherel@outlook.com

I'm a recent PhD graduate in machine learning for image and video editing. I have worked with diffusion models, VAEs, and GANs for images, videos and 3D objects. My skills range from mathematical derivations to Cuda kernel implementation. I like to work on systems that are not monolithic, but have multiple modules that require a high level of expertise. If possible, I'd prefer to work for the creative industries.

> trained from scratch

Not exactly. They mention starting from the VAE from Stable Diffusion XL and the Transformer from Phi3.

Looks like these LLMs can really be used for anything

  Location: France
  Remote: Ok
  Willing to relocate: Yes (Europe)
  Technologies: Python, PyTorch, C++, CUDA
  Résumé/CV: https://ncherel.github.io/resume.pdf
  Email: nicolas.cherel@outlook.com

I just finished my PhD in image and video processing. I have expertise in modern machine learning models (transformers, diffusion) for generation and editing. I'm looking for research positions/postdocs in creative industries.

For people asking, it's actually useful for virtual poster sessions.

Very interesting. A personal coach is certainly the most efficient way to improve. My main concern is about the legality of such a software. Even if the data is presented on screen and available, changing the UI to help the player is not accepted in all games as it can provide a strong edge. For AoE2, part of the game is about scouting and remembering key information while managing your economy.

Agreed but some questions are also a lot easier and a lot more common. In x-ray, looking for a bone fracture is a single task that requires no information about patient and can be done by an algorithm.

Surprised to read this. Having worked in the field, I see a growing interest in AI from the radiology community attested by the RSNA new AI journal. It's not about replacing radiologists but helping them in their daily work, as a safety net (double check) or as a prioritization tool.

While encouraging to think that some people will have the power to reverse climate change, it will be too late. We can't wait for the negative effects to reach rich people.

Indeed, in the full picture there must be someone paying and making this advertising profitable.

If ads were not working, I don't know where this money would go.

I agree. The problem of deepfakes is non-existing for satellite imagery as all providers are known and trusted. Not exactly the same as a random image from the internet

My favorite books of 2020 are : - Lolita by Nabokov, delightfully written and disturbing - the Foundation series by Asimov, a classic. Great plot

Would it be possible to compare image files before decompression ? For JPEG, directly compare the DCT coefficients