While 2021 was a difficult year due to the pandemic, it was also a year full of advancements in the field of AI. Louis Bouchard has compiled a great list with the best research papers that were published during the year.
2021 saw the trend of tele-working and studying remotely, prompting a number of institutions and academics to open up their research and classes to the wide public and not just their students. Examples of free content that I reported on include:
Stanford's Introduction to Robotics
Stanford's Natural Language Processing with Deep Learning
Stanford- Machine Learning with Graphs using PyTorch and NetworkX
University of Wisconsin Madison-STAT 453 Introduction to Deep Learning and Generative Models
Heinrich Heine Universität Düsseldorf - Machine Learning Lecture WS 2021/22
Memorial University - Computer Science 3200 Fall 2021 - Intro to Artificial Intelligence
Making Friends with Machine Learning - a previously internal-only Google course specially created to inspire beginners and amuse experts
You'll find all the links listed under Relevant Articles.
When going through research papers, on arXiv or otherwise, most of the time you have to go through hoops to get to the accompanying source code. For that I looked at CatalyzeX, a browser extension, available for Chrome and Firefox, that does that for you.
There's no need to worry though about Louis's list since, for each paper, he links to the related code. That aside, he's also made videos explaining the concepts behind the research. This one is for this first in the list:
The list is comprised of 38 papers:
[1] DALL·E: Zero-Shot Text-to-Image Generation from OpenAI
[2] VOGUE: Try-On by StyleGAN Interpolation Optimization
[3] Taming Transformers for High-Resolution Image Synthesis
[4] Thinking Fast And Slow in AI
[5] Automatic detection and quantification of floating marine macro-litter in aerial images
[6] ShaRF: Shape-conditioned Radiance Fields from a Single View
[7] Generative Adversarial Transformers
[8] We Asked Artificial Intelligence to Create Dating Profiles. Would You Swipe Right?
[9] Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
[10] Image GANS meet differentiable rendering for inverse graphics and interpretable 3d neural rendering
[11] Deep nets: What have they ever done for vision?
[12] Infinite Nature: Perpetual View Generation of Natural Scenes from a Single Image
[13] Portable,Self-Contained Neuroprosthetic Hand with Deep Learning-Based Finger Control
[14] Total Relighting: Learning to Relight Portraits for Background Replacement
[15] LASR: Learning Articulated Shape Reconstruction from a Monocular Video
[16] Enhancing Photorealism Enhancement
[17] DefakeHop: A Light-Weight High-Performance Deepfake Detector
[18] High-Resolution Photorealistic Image Translation in Real-Time: A Laplacian Pyramid Translation Network
[19] Barbershop: GAN-based Image Compositing using Segmentation Masks
[20] TextStyleBrush: Transfer of text aesthetics from a single example
[21] Animating Pictures with Eulerian Motion Fields
[22] CVPR 2021 Best Paper Award: GIRAFFE - Controllable Image Generation
[23] GitHub Copilot & Codex: Evaluating Large Language Models Trained on Code
[24] Apple: Recognizing People in Photos Through Private On-Device Machine Learning
[25] Image Synthesis and Editing with Stochastic Differential Equations
[26] Sketch Your Own GAN
[27] Tesla's Autopilot Explained
[28] Styleclip: Text-driven manipulation of StyleGAN imagery
[29] TimeLens: Event-based Video Frame Interpolation
[30] Diverse Generation from a Single Video Made Possible
[31] Skillful Precipitation Nowcasting using Deep Generative Models of Radar
[32] The Cocktail Fork Problem: Three-Stem Audio Separation for Real-World Soundtracks
[33] ADOP: Approximate Differentiable One-Pixel Point Rendering
[34] (Style)CLIPDraw: Coupling Content and Style in Text-to-Drawing Synthesis
[35] SwinIR: Image restoration using swin transformer
[1] DALL·E: Zero-Shot Text-to-Image Generation from OpenAI OpenAI successfully trained a network able to generate images from text captions.
[2] VOGUE: Try-On by StyleGAN Interpolation Optimization Google used a modified StyleGAN2 architecture to create an online fitting room where you can automatically try-on any pants or shirts you want using only an image of yourself.
[10] Image GANS meet differentiable rendering for inverse graphics and interpretable 3d neural rendering This promising model called GANverse3D only needs an image to create a 3D figure that can be customized and animated!
[13] Portable,Self-Contained Neuroprosthetic Hand with Deep Learning-Based Finger Control With this AI-powered nerve interface,the amputee can control a neuroprosthetic hand with life-like dexterity and intuitiveness:
[22] CVPR 2021 Best Paper Award: GIRAFFE - Controllable Image Generation Using a modified GAN architecture, they can move objects in the image without affecting the background or the other objects
[23] GitHub Copilot & Codex: Evaluating Large Language Models Trained on Code Generates Code From Words!
[26] Sketch Your Own GAN Make GANs training easier for everyone by generating Images following a sketch! Control your GAN’s outputs based on the simplest type of knowledge you could provide it: hand-drawn sketches.
You can literally spend hours going through that list watching the videos and skimming through the papers. A great list indeed.
In closing up I would like to suggest another free resource that appeared on GitHub at the end of 2021, a free book called The Deep Learning Interviews book: Hundreds of fully solved job interview questions from a wide range of key topics in AI.
Apache Geronimo Arthur has been updated with support for Common-compress, XBean, and ensures the default options are compatible with last GraalVM release.
GitHub has announced a public beta of a code scanner that automatically fixes problems. The new feature was announced back in November, but has now moved to public beta status.
