Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Autonomous AI design architect Microsoft Quiz Answers
Get Autonomous AI design architect Microsoft Quiz Answers
This learning path is a brief introduction to a new AI paradigm called Machine Teaching for Autonomous AI. Machine Teaching uses the knowledge of subject matter experts to teach AI. It integrates known control methods for stable control, Machine Learning for advance perception, and Deep Reinforcement Learning for learning strategies and human-like decision making. It’s AI integrated in industrial processes without disruption and providing real business value. It’s validated in simulation, explainable that your experts can validate. Machine Teaching for Autonomous AI stores expert operators’ skill set, enhancing and homogenizing expert operators’ best performance, and/or efficiently training novices on the job when deployed as a Decision Support tool. It helps your company achieve new levels of optimization for competitiveness, profitability, and sustainability. It provides a robust innovation path for all areas of industrial processes and business with real ROI. At the end of this learning path, you’ll be able to:
- Know the difference between automated and autonomous decision-making systems.
- Select use cases where autonomous AI will outperform both humans and automated systems.
- Leverage human expertise to design and teach AI solutions.
- Use brain design patterns to quickly design brains for any use case.
- Fill up AI specification documents that accurately describe the problem and the proposed solution.
- Use AI specification documents to talk about use cases and brain designs to diverse stakeholders.
Produced in partnership with the University of Oxford – Ajit Jaokar Artificial Intelligence: Cloud and Edge Implementations course
Prerequisites:
None
Module 1: Explore Automated Intelligence
In this module, you’ll learn about several automated methods such as math (control theory), menus (optimization algorithms) and manuals (expert systems and expert rules), which are components of Machine Teaching. In addition, you’ll learn automated intelligence strengths and limitations, when to best use these technologies and when Autonomous Intelligence methods will be the best choice.
Learning objectives:
In this module, you’ll learn how to:
- Describe automated methods included in Automated Intelligence
- Describe automated methods strengths and weaknesses
- Decide when automated methods are best used
“Produced in partnership with the University of Oxford – Ajit Jaokar Artificial Intelligence: Cloud and Edge Implementations course“
Prerequisites:
None
This module is part of these learning paths:
Quiz 1: Knowledge check
Q1. What function does the “P” term in PID control serve?
- Applies human expertise to a decision.
- Predicts the impact of a control decision.
- Prevents the controller from overshooting.
- Drives the system toward the target goal.
Q2. Which control theory technique also includes an optimization routine?
- PID
- Open Loop Control
- Feed Forward Control
- Model Predictive Control (MPC)
Q3. Which of the following techniques is NOT an Automated Intelligence technique?
- Expert Systems
- Neural Networks
- Optimization algorithms
- Control theory
Q4. How does control theory decide which is the next action to take in a system?
- Search possible options and select based on objective criteria.
- Try actions and test if they’re closer or farther from the goal.
- Calculate the next action using mathematical equations, physics, or chemistry.
- Look up options from a table.
Q5. Which of the following sentences describe strengths of optimization algorithms?
- Good for situations where you don’t have much expertise in how to control the system.
- Optimal for understanding concepts and strategies needed to complete the task.
- Generates decisions quickly with little computer resources required.
- Adapts well to fuzzy, uncertain conditions
Q6. Which automated intelligence technique would you choose for controlling the speed of the rotors on a drone?
- Control Theory
- Expert Systems
- Optimization
- Procedural Programming
Module 2: Explore Autonomous Intelligence
In this module, you’ll learn about several AI technologies such as machine learning, deep learning and deep reinforcement learning that are components of Machine Teaching. In addition, you’ll learn Autonomous Intelligence strengths, when to best use these technologies and their limitations, and when Automated Intelligence methods can support them.
Learning objectives:
In this module, you’ll learn how to:
- Describe Machine Learning for Advance Perception as a key component of Machine Teaching for Autonomous Intelligence systems
- Describe Deep Reinforcement Learning for human-like decision making as a key component of Machine Teaching for Autonomous intelligent systems
- Evaluate the advantages and limitations of Machine Learning and Deep Reinforcement Learning when used in Machine Teaching for Autonomous intelligent systems
“Produced in partnership with the University of Oxford – Ajit Jaokar Artificial Intelligence: Cloud and Edge Implementations course“
Prerequisites:
Basic knowledge of Automated Intelligence
This module is part of these learning paths:
Quiz 1: Knowledge check
Q1. Select which one of the following technologies is Autonomous Intelligence:
- Control Theory
- Expert Systems
- Deep Reinforcement Learning
- Optimization algorithms
Q2. Select which one of the following technologies has the following superpower “It automatically learns strategy”:
- Optimization algorithms
- Expert systems
- Machine Learning
- Reinforcement Learning
Q3. What is the primary strength of Machine Learning as a building block of Machine Teaching?
- Well established patterns of development towards integration on the industrial world.
- Special ability to search optimally across a huge number of options.
- Advance Perception, Prediction, and Classification.
- Outstanding technology to learn human-like strategies
Q4. What is the primary strength of Reinforcement Learning as a building block of Machine Teaching?
- Easy to benchmark and maintain.
- Searches huge number of options.
- Effective in dynamic environments.
- Only technology that can learn strategy at human level
Q5. What isn’t a weakness of Machine Learning and Deep Reinforcement Learning?
- Time and resource intensive to search and prune options when embedded on a system for industrial control.
- Requires fine-tuning of hyperparameters, especially when the conditions of the system change.
- Difficult to train, not generalizable to complex tasks that require parallel levels of abstraction.
- Black boxes with limited explainability
Module 3: Introduction to Machine Teaching
In this module, you’ll learn about capabilities and essential concepts of Machine Teaching, the steps to add intelligence into Autonomous Intelligence systems and the types of business problems that you can solve using Machine Teaching for Autonomous AI.
Learning objectives:
In this module, you’ll learn how to:
- Describe Machine Teaching
- Explain how to use Machine Teaching to outperform current industrial control systems
- Decide when Machine Teaching is the right choice for your business problem
“Produced in partnership with the University of Oxford – Ajit Jaokar Artificial Intelligence: Cloud and Edge Implementations course“
Prerequisites:
- Basic knowledge of Automated Intelligence
- Basic knowledge of Autonomous Intelligence
This module is part of these learning paths:
Quiz 1: Knowledge check
Q1. Consider the image above that shows Lunar Lander game. The goal of the Lunar Lander game is to land a spaceship in between two flags on the surface of the moon in the shortest amount of time without crashing. The landing of the spaceship must be soft and level (parallel to the ground). The spaceship sensors are: horizontal position, vertical position, horizontal velocity, vertical velocity, angle, angular velocity, left leg contact, and right leg contact. The actuators are: one main throttle under the ship and two side throttles at each side of the ship. Select the sentence that best describes the actions that you can take to control the lunar lander:
- Pitch and roll of the spaceship.
- Main throttle under the ship and two side throttles at each side of the ship.
- Move horizontally and vertically.
- Move horizontally left and right and vertically up and down.
Q2. The format of the autonomous brain has to be consistent in shape and color because:
- It looks more professional to be consistent.
- The brain design is a tool of communication, and we use the same language (shapes and color) to understand it.
- It’s mandatory.
- The shapes and colors best reflect each technology
Q3. For the Autonomous cars example, select the sentence that is FALSE:
- Actions (also known as ‘actuators’): control of steering wheel, accelerator, brake.
- Goals: arrive at destination, avoid collisions, fuel efficiency.
- Skills: staying in the lane, merging onto the highway, passing a car.
- Orchestrate skills (in order): first get into the car, then set destination, then relax and let the car drive to destination.
- Scenarios (also known as regimes): driving on highway, street roads, crowded roads, turning right, turning left
Module 4: Explore brain design patterns
In this module, you’ll learn about the main Autonomous AI brain design patterns that solve most of the Machine Teaching challenges. You’ll also learn when to best use them individually or combined.
Learning objectives:
After completing this module, you’ll be able to:
- Design Autonomous AI brains for an industrial application
- Describe the benefits and usage of brain design patterns
- Apply best practices of Machine Teaching to solve industrial business challenges
“Produced in partnership with the University of Oxford – Ajit Jaokar Artificial Intelligence: Cloud and Edge Implementations course“
Prerequisites:
- Basic knowledge of Automated Intelligence
- Basic knowledge of Autonomous Intelligence
- Basic knowledge of Machine Teaching
This module is part of these learning paths:
Quiz 1: Knowledge check
Q1. When is best to use the perception brain design pattern?
- When we need to classify images.
- When there’s an expert operator who currently decides on control strategies based on advance perception.
- When we need to predict product demand.
- When we need to identify sounds
Q2. When is best to use the functional brain design pattern?
- When there are different tasks that can only be controlled by one expert operator at a time.
- When there are different tasks that are executed using the same control actions.
- When there are different tasks that are executed using independent control actions.
- When there’s an expert operator currently performing the task.
Q3. When is best to use the strategy brain design pattern?
- When we identify strategies that are controlled using independent control actions.
- When there are different tasks that are controlled by different expert operators independently.
- When there’s an expert operator currently performing the task.
- When we identify strategies that are controlled using the same control actions.
Module 5: Explore the art of selecting use cases for winning Autonomous AI
In this module, you’ll learn how to identify use cases that are a good fit to be solved using Machine Teaching for Autonomous AI.
Learning objectives:
After completing this module, you’ll be able to:
- Select use cases for Autonomous AI
- Evaluate scenarios where you can use Autonomous AI
“Produced in partnership with the University of Oxford – Ajit Jaokar Artificial Intelligence: Cloud and Edge Implementations course“
Prerequisites:
- Basic knowledge of Automated Intelligence
- Basic knowledge of Autonomous Intelligence
- Basic knowledge of Machine Teaching
- Basic knowledge of brain design patterns
This module is part of these learning paths:
Quiz 1: Knowledge check
Q1. The purpose of the Heating, Ventilation and Air Conditioning (HVAC) use case is to regulate the damper to minimize energy cost while keeping temperature within comfort (as dictated by a set point) and keeping CO2 within legal limits. The current methods include a PID controller with fixed automated timer. Select which one of the following limitations of the current methods does NOT apply:
- Dynamic highly variable systems.
- Competing optimization goals or strategies.
- Unforeseen starting or system conditions.
- Dependency on human operators
Q2. MineCo Is a company that mines precious metals from ore. The first and roughest stage of ore processing is crushing the ore, in this case with a gyratory crusher. Operators manually control supervisory settings for the gyratory crusher. Low-level control is done with PID controllers. The goal is to maximize throughput while also maximizing efficiency. The size and hardness of the incoming rocks is variable within a set of well-known limits. Select which one of the following limitations of the current methods does NOT apply:
- Dynamic highly variable systems.
- Competing optimization goals or strategies.
- Unforeseen starting or system conditions.
- Independency on human operators
Q3. A logistics company has a large warehouse with many front doors to load trucks for daily distribution to multiple stores. Wherever you locate the goods in the warehouse will influence how easy or difficult it’s to access them and how fast a truck can be loaded. For perishable produce, like lettuce, that needs to arrive at the destination fast you want to have it distributed in every door to facilitate its transportation. For long-lasting food, like potatoes or onions, it can be concentrated in one location where you can get it when you need it. The goal is to maximize throughput and minimize truck loading time. Current methods include some optimization algorithms and the experience of the scheduler operator. Select which of the following limitations of the current methods don’t apply:
- Dynamic highly variable systems.
- Competing optimization goals or strategies.
- Unforeseen starting or system conditions
- Dependency on human operators
Module 6: Explore AI Specification document
In this module, you’ll learn how to fill out an AI specification document for an Autonomous AI use case.
Learning objectives:
After completing this module, you’ll be able to:
- Understand how to fill out an AI Specification document for a use case of your choice
Prerequisites:
- Basic knowledge of Automated Intelligence
- Basic knowledge of Autonomous Intelligence
- Basic knowledge of Machine Teaching
- Basic knowledge of brain design patterns
- Basic knowledge of selecting use cases for Autonomous AI
This module is part of these learning paths:
Quiz 1: Knowledge check
Q1. In the Moab sample, what are pitch and roll?
- Environment states.
- Control actions.
- Goals.
- Strategies
Q2. What does it mean that the training episode length is 250 iterations?
- Is the maximum number of actions the brain will be allowed to take to succeed.
- Every episode has exactly 250 iterations.
- Even if the brain fails before 250 iterations, the number of iterations per episode will be 250.
- It means that the brain will never face more than 250 iterations during training nor deployment
Q3. (Select the most correct answer) The Moab brain design presented on the AI spec in this module is an example of:
- A modular brain design.
- A monolithic brain design.
- A monolithic brain design with an advanced perception preprocessor.
- A modular brain design with an advanced perception preprocessor.
Q4. What is the difference between environment states and configurable scenarios?
- Environment states are the sensors or input to the brain and configurable scenarios are the lessons given to the brain to practice controlling the device.
- Environment states are the control actions performed by the brain at each iteration.
- Environment states are the input to the brain and configurable scenarios are the output of the brain.
- Environment states and configurable scenarios are the same
Q5. What type of simulator is implemented in the Project Moab sample?
- A subrogate simulator.
- A Data Driven Simulator (DDS).
- A first-principles simulator.
- The Moab doesn’t have a simulator
Conclusion:
I hope this Autonomous AI design architect Microsoft Quiz Answers would be useful for you to learn something new from this problem. If it helped you then don’t forget to bookmark our site for more Coding Solutions.
This Problem is intended for audiences of all experiences who are interested in learning about Data Science in a business context; there are no prerequisites.
Keep Learning!
