TRENDING ON ONEINDIA
-
"Prevent Black Day," Says Retired HC Judge -
Owning A Maruti Suzuki Is Now A Special Occasion — Here's How -
Mobiles Republic Day 2019 — Discounts On Best Smartphones -
From Kohli To Dhoni; Five Reasons Behind India's Win In 2nd ODI -
Every Place You Need To Visit If You Are A First-Timer In Una -
Strange Case Of A Woman Who Cannot Hear Male Voices -
Company Fixed Deposits In India Which Offer Yields Of Near 10% -
Emraan Justifies Himself After Calling Aishwarya 'Plastic'
Two AI researcher from Google recently detailed their creation of an advanced offline AI for text classification. The AI is also capable of running on low-end smartphones and achieved 86.7% accuracy on a simple data set, while 83.1% on a more complex and multifaceted data set.
The tests were done locally, as well as on-the-fly training. The system uses self-governing neural networks to keep the footprint low and eliminate the need for the cloud. While the accuracy and training are helped along by loss-defined ground truth, the model's base is modified by an algorithmic tool.
This is basically a form of supervised learning that doesn't require major supervision from humans and can teach an AI quickly to learn on its own. Here, the AI classified blubs of text, categorizing them on the basis of whether they may be a sentence, a phrase, a maths equation, an identification number, or something else.
Since machine learning and neural networking features are rapidly becoming a norm for flagship smartphones, an AI like this could pave the way for lower-end devices with AI capabilities. Text classification might not sound like a big deal, but it sure does has the potential when it comes to things like live translation, UI navigation, and projects aiming to help the differently-abled navigate their devices.
In places where there's a lack of internet connection and hefty hardware, this solution could prove to be handy.
Besides, Google's study into the use of AI-based Lymph Node Assistant (LYNA) shows how artificial intelligence can come in handy in real life scenarios. In its AI blog, the company details about helping human pathologists diagnose metastatic breast cancer in lymph node slides.
The use of LYNA makes it easy to detect cancer cells, helping pathologists diagnose them. In fact, Google says it cut the review time by half for each slide used. The new study is a continuation of the research on the use of AI for detecting and diagnosing cancer which was reported early this year. Earlier, Google also showed how its DeppMind is capable of helping pathologists to detect cancer cells.
