0% found this document useful (0 votes)
6 views2 pages

Class 9 AI Deployment Notes

Deployment refers to making AI models available for use, with three main types: On-Device/Premise, Cloud, and Edge Deployment. Each method has its pros and cons, such as accessibility and data privacy concerns for Cloud Deployment, while On-Premise Deployment offers local data processing but higher costs. Edge Deployment provides portability and immediate analysis but may face limitations in processing power and battery life.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
6 views2 pages

Class 9 AI Deployment Notes

Deployment refers to making AI models available for use, with three main types: On-Device/Premise, Cloud, and Edge Deployment. Each method has its pros and cons, such as accessibility and data privacy concerns for Cloud Deployment, while On-Premise Deployment offers local data processing but higher costs. Edge Deployment provides portability and immediate analysis but may face limitations in processing power and battery life.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

Deployment

Deployment is the mechanism through which the developed and trained


AI model is made available to users for actual use.

(I) On-Device/Premise Deployment

This is the case when the actual device at the point of use (e.g., your
phone or Smartwatch) directly houses or stores the AI model. It makes it
very easy to use it, because you can work even when you’re not
connected to the Internet.

(II) Cloud Deployment

AI models can occasionally be so large and complex that they require a lot
of processing power to execute. For such scalable and flexible AI models,
deploying them on the device is not feasible. So, these are deployed on
the cloud. As you must be aware, a cloud means the storage, databases,
and processing systems available on the Internet servers. There are many
clouds, some of which are by companies like Microsoft, Google, and
Amazon.

(III) Edge Deployment

In this deployment, the deployment happens on edge devices. Edge


devices, such as computers or intelligent devices that enable real-time
raw data collection and processing for quicker, more accurate analysis,
e.g., smart sensors of IoT.

(e.g., AI cameras on traffic lights can differentiate vehicles and people


from surroundings, or you must have seen or heard about the robots that
move around without crashing?)

Hybrid deployment involves the use of more than one deployment


technique.

1. Cloud Deployment

In this method, the AI model would be hosted on a cloud platform, and


healthcare providers or patients could upload retinal images securely for
analysis.

Pros

Easy to update and maintain the Al model centrally.

Accessible from anywhere with an internet connection.

Cost-effective, as infrastructure costs are shared.

Cons

Requires a stable internet connection, which may be challenging in remote


areas.

1
Data privacy and security concerns with transmitting sensitive medical
data.

2. On-premise Deployment

With this approach, the AI model would be installed locally at healthcare


facilities or clinics, running on their own servers or computers.

Pros

No need for an internet connection, suitable for remote areas.

Data remains on-site, addressing privacy and security concerns.

Faster analysis, as data doesn't need to be transmitted.

Cons

Higher upfront costs for each location to acquire the necessary hardware.

Maintenance and updates become more complex across multiple sites.

Potential for inconsistent performance if hardware specifications vary.

3. Edge Device Deployment

In this method, the AI model would be deployed on edge devices like


smartphones, tablets, or specialized medical devices with built-in
cameras.

Pros

Highly portable and accessible, even in remote areas.

No need for an internet connection or specialized hardware.

Immediate analysis and feedback for patients.

Cons

Limited processing power on edge devices may impact model


performance.

Battery life and storage constraints on mobile devices.

Securing and updating models on numerous devices can be challenging.

You might also like