Prebuilt models, as the name suggests, are ready-made models proffered by Microsoft. These models are designed to understand the common entities or actions in most general applications. They include Key Phrase Extraction, Language Detection, Named Entity Recognition, Sentiment Analysis, and Text Recognition. Microsoft’s prebuilt models are superb at addressing common tasks but may not handle specific business needs decisively.
Custom models, on the other hand, are designed and tailored by you or your business users to meet specific business requirements. They are crafted to understand and analyze specific data unique to the needs of your business. For instance, you may want to create custom models that analyze billing details or customer feedback in your unique style.
Comparison
Let’s delve into some specifics for effective understanding:
- Training Data: Pre-built models come already trained on general data and are ready to use ‘out of the box’. They may not require any specific business data to function. Custom models, however, take your personal or company’s business data to train and perform tasks. They are programmed to capture specific intents and entities, making them more flexible than prebuilt models.
- Use Cases: Prebuilt models are designed for general use-cases like analyzing sentiment, entities, and detecting languages in a text. Custom models, however, cater to user-specific requirements like analyzing certain phrases in customer feedback or adding particular attributes to their products.
- Integration: Prebuilt models can be readily integrated with apps, requiring little or no modification. Custom models, conversely, require extensive code modifications to align them accurately with business apps.
- Accuracy: While prebuilt models may deliver decent accuracy on general datasets, their efficacy dwindles with industry-specific data sets. Custom models trump in this context as they can be tailored to deliver more precise outcomes on industry-specific datasets.
- Maintenance: Prebuilt models require minimal maintenance since they are built and managed by Microsoft. Contrarily, custom models require continuous training and updating to stay effective.
Advantages of Prebuilt and Custom Models
Prebuilt Models:
- Simplified usage: Prebuilt models provide a quicker and simpler way to supplement your apps with AI capabilities without getting entrenched in coding.
- Saves Time: They save developers a significant amount of time since they do not require training.
Custom Models:
- Specific: Since these models are trained on your data, they bring more relevance and accuracy to your application.
- Flexible: They offer immense flexibility to tailor your AI applications to your unique business needs.
Overall, the choice between prebuilt and custom models relies heavily on your specific business requirements. It is advisable to use prebuilt models for generic tasks while reverting to custom models to handle more personalized, business-specific tasks.
Practice Test
True/False: Prebuilt models are ready-made solutions, while custom models are adaptable to specific business needs.
- True
- False
Answer: True
Explanation: Prebuilt models are created by Microsoft and ready for use, components such as the Vision AI Model can be used as they are. Custom models, however, are built and trained based on a specific business scenario, and can therefore be fully adapted to specific needs.
Which of the following are differences between prebuilt models and custom models? (Multi-select)
- A. Level of customization
- B. Training data
- C. Cost
- D. Speed of implementation
Answer: A, B, C, D
Explanation: All provided options highlight core differences between prebuilt and custom models in Microsoft Power Platform.
True/False: When using custom models, there is no need for initial training data.
- True
- False
Answer: False
Explanation: To build a custom model, initial training data is necessary as it is used to teach the model how to handle similar data in the future.
In which scenario would you typically use a prebuilt model? (Single select)
- A. When you have a standard task
- B. When you have a complex, company-specific task
- C. When you need a quick, cost-effective solution
- D. All of the above
Answer: A, C
Explanation: Prebuilt models are typically used for standard tasks or when a quick and cost-effective solution is needed because they are readily available and do not require substantial modifications.
True/False: Custom models are typically faster to implement than prebuilt models.
- True
- False
Answer: False
Explanation: Custom models, unlike prebuilt models, require more time to develop and train. Therefore, prebuilt models are typically faster to implement.
What’s a main advantage of using a custom model? (Single select)
- A. Requires less time to develop
- B. Is inexpensive to create
- C. Can handle very specific tasks
- D. Can be used unmodified
Answer: C
Explanation: The main advantage of a custom model is its ability to handle very specific tasks, as it is trained on company-specific data and designed to perform specific functions.
True/False: Prebuilt models generally have a higher level of accuracy than custom models.
- True
- False
Answer: False
Explanation: This statement is not necessarily true. The accuracy of a model depends on the quality and quantity of training data. In some cases, a well-trained custom model may have higher accuracy than a prebuilt model.
True/False: Custom models require a larger amount of data to train as compared to prebuilt models.
- True
- False
Answer: True
Explanation: Since custom models are built to accommodate a company’s specific needs and scenarios, they require more data for training to increase their accuracy and versatility.
Which of the following is the biggest disadvantage of using a prebuilt model? (Single select)
- A. Limited customization
- B. Cost
- C. Speed of implementation
- D. Training data is too complex
Answer: A
Explanation: The biggest disadvantage of prebuilt models is their limited potential for customization as they are designed to handle general tasks and not specific business needs.
True/False: Prebuilt models require special skills to use.
- True
- False
Answer: False
Explanation: Prebuilt models are designed by Microsoft to be ready-made, straightforward, and user-friendly. Therefore, they don’t require special skills to utilize.
Interview Questions
What are prebuilt models in the context of the Microsoft Power Platform?
Prebuilt models are AI models that are ready to use and have been pretrained on a large body of data so that you don’t need to train them yourself. They are designed to perform specific tasks like language detection, sentiment analysis, and image recognition.
How are custom models different from prebuilt models?
Custom models are built and trained by the users in the context of their specific needs and data. They offer greater flexibility and precision as they are optimised for a specific use case or data set.
Why would someone use a prebuilt model instead of a custom model?
Prebuilt models save time as they are already trained and ready to use. They are also simple and require less expertise to set up and consume.
Can you give an example of a task where you might use a custom model?
A custom model can be used in predicting customer churn, diagnosing diseases, or any complex task that needs a specific set of training data which is unique to your business context or sector.
What are the disadvantages of using prebuilt models?
The disadvantages include less accuracy for specific applications, lack of flexibility, limited ability to account for unique characteristics in your data, and potential bias in the training data.
What data is needed to create a custom model in Microsoft Power Platform?
To create a custom model, you need training data that includes both input variables and corresponding output. The specificity and variety of this data will determine the efficacy of your model.
Can you modify prebuilt models according to your unique requirements?
No, prebuilt models cannot be modified as they are trained on a set of data that Microsoft supplies, and are designed for generalised tasks.
How does using a custom model benefit in terms of accuracy?
Custom models can be more accurate as they are trained on your specific data set and are made to address a specific problem or task, hence they can offer better prediction accuracy.
What skills are required to build a custom model?
To build a custom model, one needs understanding of machine learning principles, knowledge of data science and experience in programming languages such as Python or R.
In what scenarios is it more beneficial to use prebuilt models?
In scenarios where you don’t have the time, data, or expertise to build and train a custom model, prebuilt models come in handy. They’re useful for general tasks without requiring extensive setup or fine-tuning.
Is knowledge in machine learning required to use prebuilt models in the Microsoft Power Platform?
No, prebuilt models require no specific knowledge in machine learning to use. They are designed to be straightforward and user-friendly, even for those without AI expertise.
Can custom models in Microsoft Power Platform learn and improve over time?
Yes, you can train your custom model on new data over time to improve its performance or to ensure it keeps up with changes in patterns in your data.
How are prebuilt models maintained in Microsoft Power Platform?
With prebuilt models, the maintenance is handled by Microsoft. Users do not have to worry about updating or maintaining these models.
Can prebuilt models deliver insights directly without any processing?
Yes, prebuilt models can deliver insights directly as they are ready-to-use models pre-trained on diverse data sets.
Can the training of custom models in the Microsoft Power Platform be automated?
Yes, Power Platform provides the capability to automate the training process of custom models using AI Builder.
extutorials.com