How AI is Cracking a Hard Problem

WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

The machine's olfaction starts with sensors that detect and identify molecules in the air. (Representative)

100 years ago, Alexander Graham Bell asked National Geographic readers to do something bold and fresh — “to found a new science.” He pointed out that science already exists based on the measurement of sound and light. But there was no science to odor. Bell told his readers to “measure the smell.”

Today, the smartphones in most people's pockets offer impressive built-in capabilities based on sound and light science: voice assistants, face recognition and image enhancement. Odor science compares nothing. But that situation is changing, as advances in machine spinning, also known as “digitized summaries,” are finally answering Bell's call to action.

Research on mechanical olfaction faces a formidable challenge due to the complexity of the human sense of smell. While human vision relies primarily on receptor cells in the retina—rods and three types of cones—smell is experienced through about 400 types of receptor cells in the nose.

The machine's olfaction starts with sensors that detect and identify molecules in the air. These sensors work just like the receptors in your nose.

But to be useful for people, mechanized scents need to go a step further. The system needs to know how a particular molecule or set of molecules smells to a human. For this, machine olfaction requires machine learning.

Applying machine learning to smell

Machine learning, and specifically a type of machine learning called deep learning, is at the heart of notable developments such as voice assistants and facial recognition apps.

Machine learning is also key to digitizing smell because it can learn to map the molecular structure of an odorant compound to textual odor descriptors. The machine learning model learns words that humans use — for example, “sweet” and “sweet” — to describe when they encounter specific odor-causing compounds, such as vanillin.

However, machine learning requires large data sets. There is an unimaginably large amount of audio, image and video content on the web that can be used to train artificial intelligence systems that can recognize sounds and images. But machine decay has long been plagued by a lack of data, in part because most people can't verbally describe smells as easily and identifiably as they can describe sights and sounds. can do Without access to web-scale datasets, researchers were unable to train truly powerful machine learning models.

However, things began to change in 2015 when researchers launched the DREAM Olfaction Prediction Challenge. The competition released data collected by Andreas Keller and Leslie Voshall, biologists who study olfaction, and invited teams from around the world to submit their machine learning models. The models were supposed to predict odor labels such as “sweet,” “floral,” or “fruity” for odor-causing compounds based on their molecular structure.

The top-performing models were published in a 2017 paper in the journal Science. A classic machine learning technique called random forest, which combines the output of multiple decision tree flowcharts, emerged as the winner.

I am a machine learning researcher with a long-standing interest in applying machine learning to chemistry and psychiatry. The Dream Challenge piqued my interest. I also felt a personal connection to Zalf. My family traces back to the small town of Kanauj in northern India, the perfume capital of India. Moreover, my father is a chemist who spent most of his career analyzing geological samples. Machine scents thus provide an irresistible opportunity at the intersection of perfumery, culture, chemistry and machine learning.

After the Dream Challenge ended, developments in machine rotation began to pick up steam. During the COVID-19 pandemic, many cases of smell blindness, or anosmia, were reported. The sense of smell, which usually takes a back seat, emerged into the public consciousness. Additionally, a research project, the Pyrfume Project, made increasingly large datasets publicly available.

Deep scent.

By 2019, the largest datasets had grown from less than 500 molecules in the DREAM challenge to nearly 5,000 molecules. A research team at Google led by Alexander Viltsko has finally succeeded in revolutionizing deep learning in machines. Their model, based on a type of deep learning called a graph neural network, provides state-of-the-art results in machine learning. Wiltschko is now the founder and CEO of Osmo, whose mission is to “give computers a sense of smell.”

Recently, Wiltschko and his team used a graph neural network to create a “principal odor map,” where perceptually similar odors are placed closer together rather than different ones. It wasn't easy: small changes in molecular structure could lead to large changes in olfactory perception. Conversely, two molecules with very different molecular structures can still smell almost exactly the same.

Such advances in breaking the code of smell are not only intellectually exciting, but also have very promising applications, including personalized perfumes and fragrances, improved insect repellents, new chemical sensors, disease detection, and more. Includes faster detection, and more realistic augmented reality experiences. The future of machine tools looks bright. It also promises to smell good.

(the author:(Amboj Tiwari, Professor of Statistics, University of Michigan)

(Disclosure Statement:Amboj Tiwari does not work for, consult with, participate in, or receive funding from any company or organization that benefits from this article, and has no related affiliations beyond his academic appointment. not disclosed)

This article is republished from The Conversation under a Creative Commons license. Read the original article.

(Other than the headline, this story has not been edited by NDTV staff and is published from a syndicated feed.)

WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

Leave a Comment