This week in AI & Machine Learning: Andrew Ng calls for data-centric AI and Plainsight’s CEO discusses the hunt for world-class AI talent.
A Note from the Author
Realizing the impressive powers of AI is only possible with the right team, a savvy group leveraging a diverse set of skills and experiences. Organizations know they need such teams and they’re making hefty investments to find them. Still, this tends to prove easier said than done. Plainsight’s Co-Founder and CEO, Carlos Anchia, joins a range of AI industry leaders in discussing the types of roles organizations should hire for and the types of professionals they should look for. Check it out on CIO.
Artificial Intelligence News
An Argument for Data-Centric AI
AI is only as good as the data used to train it. That was effectively the argument Google Brain founder and AI thought leader Andrew Ng made when he spoke at the EmTech Digital Conference in late March. Ng advocates for a ‘data-centric’ approach to AI, one that’s highly disciplined in its emphasis on quality and consistency. Data-centric AI, Ng suggests, could be the key to improving adoption in industries like healthcare and manufacturing that have been comparatively slow to adopt new solutions. Read a detailed summary of Ng’s address and his ongoing work to promote data-centric AI.
Google’s New Pre-Training Framework for Multimodal Video Captioning
Multimodal video captioning presents a pair of challenges, as models must process and understand input videos while generating accurate captions. Google researchers have introduced a new pre-training framework called multimodal generative pre-training (MV-GPT) that has achieved impressive results for both captioning and understanding videos. Models are trained with two separate generation loss processes, called ‘forward’ and ‘backward’: Forward generation sees the model create a future utterance based on input frames and the present utterance. In backward generation, the model uses future utterances to generate present utterances. Learn more on Google’s blog.
AI for Assessing Soybean Crops
A new study published in Computer and Electronics in Agriculture shows how researchers developed a new convolutional neural network capable of accurately identifying defoliation in drone imagery of soybean crops. It is far superior to existing methods at both identifying instances of crop defoliation and avoiding incorrect labels. The study’s authors hope that DefoNet, trained on nearly 100,000 images, can provide for better decision making in efforts to mitigate and manage crop loss. Read an overview of the study and visit Plainsight’s blog to learn more about vision AI use cases for agribusinesses.
Join our Community
See you next week! Until then, keep the conversation going on Plainsight’s AI for All Slack Channel.
About the Author & Plainsight
Bennett Glace is a B2B technology content writer and cinephile from Philadelphia. He helps Plainsight in its mission to make vision AI accessible to entire enterprise teams.
Plainsight’s vision AI platform streamlines and optimizes the full computer vision lifecycle. From data annotation through deployment, customers can quickly create and successfully operationalize their own vision AI applications to solve highly diverse business challenges.