This week in AI & Machine Learning: building better AI assistants, tiling for better computer vision, trends in reinforcement learning, and more!
Artificial Intelligence News:
Project CAIRaoke: Building the assistants of the future with breakthroughs in conversational AI
Meta announces project CAIRaoke, a breakthrough in conversation AI. This announcement in research comes as part of Meta’s initiative to bring more AI power into the metaverse. Read the technical details on the Meta AI blog or for more of a summary, check out this post from the BBC.
We’re excited to share details on Project CAIRaoke, a breakthrough in conversational AI. With this end-to-end system, we’ll be able to have much more personal, contextual conversations than we can with the systems people are familiar with today. https://t.co/BXNffFN6Z8 pic.twitter.com/YrqnI7G7Zo— Meta AI (@MetaAI) February 23, 2022
Tiling in Computer Vision: The Key to Small Object Detection
In our latest blog, we explain why tiling is an important technique used in computer vision for detecting small objects in high-resolution images. Read more on the official Plainsight blog.
Trends in Deep Reinforcement Learning with Kamyar Azizzadenesheli
In this episode of TWiML, Kamyar Azizzadenesheli discusses the trends in reinforcement learning and why there is a perception of development slowing.
A majority of consumers believe AI will positively impact the next decade.
A new study suggests that most consumers feel like AI positively impacts products and are even happy to tune preferences if it improves the experience. Read the complete analysis on VentureBeat.
Probing Image-Language Transformers for Verb Understanding
Google’s DeepMind asks whether multimodal transformers can perform well on verb translation to images. Read the results on the official DeepMind blog.
Multimodal transformers perform well on vision and language tasks like visual question answering, but do they understand verbs? Our team collects the SVO-Probes dataset to find out! https://t.co/fcWvrWQnOH— DeepMind (@DeepMind) February 23, 2022
Work by Lisa Anne Hendricks & @aidanematzadeh 1/ pic.twitter.com/gRNf4XS3TO
Can Machine Learning Models Overcome Biased Datasets?
Adam Zewe, from the MIT News Office, dives into why machine learning models can be biased and highlights potential solutions for overcoming it. Read the post at MIT News.
I’ll see you next week! You can join the discussion on the AI for ALL Slack.