Try all of the on-demand periods from the Clever Safety Summit here.

Andrew Ng’s cloud-based platform for pc imaginative and prescient, Landing AI, is taking over the arrival of synthetic intelligence (AI) improvement amongst firms of all sizes with its newest providing, LandingLens. The answer guarantees to facilitate swift creation and testing of pc imaginative and prescient AI tasks, with out the necessity for intricate programming abilities or prior AI expertise. 

“We began by exploring the manufacturing sector, one of many hardest industries through which to deploy pc imaginative and prescient. Then we discovered the instruments we had constructed for manufacturing, with comparatively few modifications, can be helpful for a lot of different pc imaginative and prescient functions,” mentioned Ng, famous AI tutorial, and founder and CEO of Touchdown AI. 

The corporate introduced right now that its flagship pc imaginative and prescient product, LandingLens, is now obtainable for a free trial, coupled with a brand new pricing scheme that permits pay-as-you-go utilization past the preliminary trial interval.

“With the brand new platform, we goal to broaden our device’s use circumstances throughout a number of different industries,” Ng advised VentureBeat. “To me, it’s about attaining our aim of democratizing the creation of AI.”


Clever Safety Summit On-Demand

Study the vital function of AI & ML in cybersecurity and business particular case research. Watch on-demand periods right now.

Watch Here

“We would like everybody to start out without spending a dime and check out it out to grasp its use circumstances. We’re desirous to make it obtainable for extra folks,” he mentioned. 

A knowledge-centric method to AI and pc imaginative and prescient

In response to Ng, the platform’s data-centric AI system focuses on knowledge as an alternative of code, and as numerous industries more and more embrace AI options, a elementary shift is important to unlock the entire potential of this expertise. 

LandingLens prioritizes enhancing knowledge high quality for AI fashions, thereby enabling its performance, even in circumstances the place firms have restricted knowledge obtainable for coaching the AI fashions, a typical problem encountered by most corporations. The “data-centric” technique includes coaching AI fashions to perform proficiently with modest quantities of high quality knowledge relatively than relying solely on the huge datasets that usually underpin AI functions in large-scale web firms.

“Over the previous few years, we did a lot work with clients that always had small datasets. Throughout these experiences, we found a number of expertise steps and optimizations that now allow our algorithm to work properly on smaller datasets,” mentioned Ng.

He defined that the mannequin was skilled on a ResNet dataset for picture recognition, and within the backend, LandingLens’s pretrained algorithm makes use of AI-based automated hyperparameter tuning, enabling it to work properly with datasets of each dimension. When knowledge is handed by way of the mannequin, it’s optimized by way of quite a few steps to ship well-analyzed, high-quality output and detailed insights. 

Not too long ago, therapeutic antibody discovery agency OmniAb used LandingLens to efficiently automate its visible inspection course of, considerably growing effectivity and throughput. As well as, the platform aided OmniAb in growing AI entry inside its group to be used circumstances that contain people who find themselves not high-level scientists. 

How does it work? 

To take care of knowledge consistency inside LandingLens, the platform makes use of a sophisticated labeling expertise that routinely detects and corrects mislabeled pictures, enhancing total knowledge high quality. 

This collaborative labeling method permits a number of customers to label pictures and facilitates the method of reaching a consensus by way of knowledge cloud and edge system deployment capabilities. In consequence, deploying and testing your mannequin could be achieved with only a few clicks of the mouse. Customers can choose the deployment choice that most closely fits their necessities, starting from a home windows software to a programmatic API.

Moreover, LandingLens employs a continuous-learning mechanism that ensures that the created mannequin stays updated by integrating new knowledge from the deployment surroundings to retrain the mannequin.

“We wish to make the mannequin improvement workflow straightforward for customers. The normal method to creating AI fashions has all the time been labeling, coaching to deployment. We wish to ease this improvement workflow by having customers not write a lot code, however focus extra on knowledge entry,” added Ng. 

Picture supply: Touchdown AI.

Touchdown AI’s future focus on pc imaginative and prescient

Ng mentioned the corporate would proceed to give attention to creating the LandingLens platform as a single device that serves a number of pc imaginative and prescient functions.

“Use circumstances in pc imaginative and prescient are at the moment protecting us very busy. Many purchasers throughout industries are requesting us so as to add extra options for circumstances comparable to streamlining heterogeneous knowledge. So our present roadmap includes much more work to do in pc imaginative and prescient,” mentioned Ng. 

By means of the LandingLens platform, Ng goals to resolve points discovered right now with customization or longtail AI mannequin improvement, which he sees as essentially the most vital barrier to widespread AI adoption. 

“The one manner for organizations to unlock most worth from their AI tasks is once they have the freedom to customise their AI system as they want. They will do that by engineering the info relatively than the code. This fashion, firms can regulate to the shifting market necessities and develop higher fashions utilizing lesser human assets,” defined Ng. “So, I’m enthusiastic about facilitating the aim of additional democratizing entry to AI creation.”

The corporate is pursuing functions in automotive, electronics and medical system manufacturing sectors. Ng mentioned embracing a data-centric AI methodology and implementing AI and deep learning-based options for pc imaginative and prescient eventualities will profit this various vary of industries.

Source link