We’ll examine the algorithms further in order to grasp how the tools integrate into interaction processes. The latter are MDP, which assist a robot in making choices concerning activities. The second category is HMM, which enable the robot to ‘vision’ places which it was unable to see. We’ll also explore why probabilities are key in helping robots plan paths safely and accurately. Ready to see how robots make their moves? Let’s explore Markov models!

How Do Markov Models Work?

Markov models work through a series of steps, using probabilities to help a robot guess what might happen next. This is essential for robots that need to navigate tricky paths or change their plans quickly, especially when they can’t “see” everything around them.

With Markov models, robots can “learn” from past moves and make decisions, even when they don’t have all the information—just like how we might decide to turn down a road based on signs, even if we can’t see the destination.

Markov Decision Processes in Robotics

When it comes to making decisions, robots can’t depend on instinct like we do (this is a big difference). Instead, they use Markov Decision Processes (MDP) which are similar to decision trees that guide their actions step-by-step. However, these processes work differently, because they calculate the best possible choice at each moment. Although robots are smart, they lack the intuition that humans have.

What Are Markov Decision Processes (MDPs)?

MDP represents a math model of making decisions where some state transitions are stochastic (due to random events — such as rolling the dice) and/or controlled by an agent (in our case, robot). While this does simplify things, it allows the robot to determine what action should be performed in any situation. This is done by looking at its current environment and its goals.

MDPs consist of four key components:

• States (S): These represent all the possible situations a robot can be in. For example, a robot could be in a room or performing a task.
• Actions (A): The robot has several choices it can make: it can move forward, turn left, or (even) pick up an object.
• Transitions (T): These describe how actions affect the robot’s state. For example, moving forward might change the robot’s location from one room to another.
• Rewards (R): The robot receives rewards for taking specific actions. A higher reward indicates a more desirable action or outcome, such as completing a task successfully.

How MDPs Help Robots Make Decisions

However, robots in the real world often have to deal with uncertainty. They cannot know precisely the outcomes of when they do one action or another. This is resolved with MDPs, which factor in the probabilistic nature of a robot decision-making process i.e. which actions are likely to lead to certain results?

MDPs take into account the different probabilities of some states and all actions in them to allow for optimal decision making by a robot in order to maximize rewards over long time horizons. Example — In a maze robot may try to choose between different paths and check the path that gives it highest probability of reaching its goal.

Real-World Applications of MDPs in Robotics

MDPs are popular for tasks such as robot path planning and robotic navigation. Following this word of caution, here are some pragmatic instances

• Autonomous Vehicles: MDPs allow self-driving cars to make safe choices on the road by analyzing every decision (to turn, stop or accelerate) under the present state of traffic so they remain in control at all times.
• Delivery Robots: To find the optimal way of delivering packages while conserving energy and avoiding obstacles, these robots use Partially Observable Markov decision processes (MDPs).
• Robotic Arms: MDPs are used to teach robots on an assembly line, allowing them to identify which tool and motion chain should be employed to accomplish a specific task in a precise manner.

Hidden Markov Models for Robot Navigation

A Hidden Markov Model (HMM) is a special case of the Markov model. A Markov model, however, assumes that the current state is fully visible to the observer (or agent), whereas an HMM tackles situations where the state is hidden/missing/partially observable. In other words, the robot may not always know its exact position or environment due to incomplete sensors or uncertain conditions.

In an HMM, the robot’s state is hidden, meaning it can only observe outputs that give it indirect information about the environment. These observations are used to make inferences about its current state, improving its ability to navigate and make decisions.

How HMMs Improve Robot Navigation

Hidden Markov Models are one of the best choice to use in the case of robot navigation when dealing with observation uncertainty or degree of failure as the sensor data are not sufficient to get all movements required for successful movement. This could be due to environmental factors, such as a robot walking in a dark room where there may not always be visual inputs available.

Using an HMM, it can interpret sensor readings (like motion or sound) and estimate its hidden position.

HMMs help robots:

• Estimate unknown states: By interpreting sensor data, robots can make educated guesses about where they are, even if they can’t see everything clearly.
• Adapt to changing environments: Robots can adjust their behavior based on updated information, such as moving faster if they detect an open path.
• Handle uncertainty: By considering the probability of different outcomes, robots can make decisions despite the unknowns in their environment.

Applications of HMMs in Robotics

HMMs are commonly used in:

• Robots and Speech Recognition: Ever imagined giving your robot voice commands? Robot voice command capabilities rely on HMMs to understand spoken instructions, especially when the audio is noisy or pronunciations are not clear.
• Detection and mapping: In cases where GPS or other sensors are not available for localization of the robot, HMMs may be used to estimate its location.
• Object Tracking (Evaluating moving objects): By using cameras and other sensors, robots can track the direction of movement for any generally available article in nature utilizing HMMs.

Probabilistic Models in Autonomous Robotics

Probabilistic models in robotics are tools that can assist robots and robot designers in making decisions or choices when faced with uncertainty and/or randomness of sensory information. While in deterministic models the robot applicably consistently acts a certain way, hence always having her future defined and solved for; probabilistic models accept that not all minutes can be foretold.

These models help robots reason about:

• Sensor data uncertainty: Robots cannot be sure 100% what their sensors are sensing due to all sorts of that comes into play, and so yes probabilistic models help factor those chances.
• Action randomness: The result of a robot action is not always determinable, but probabilistic models can predict the expected outcome.
• Environmental changes: The robot may not be aware and does not know how its environment is changing, but with probabilistic models it can prepare for any case.

Why Probabilistic Models Matter for Autonomous Robots

In autonomous robotics, where robots must function without human intervention, probabilistic models are crucial because they allow robots to:

• Make decisions under uncertainty: Robots can handle situations where sensor data is noisy or incomplete, and they can still take appropriate actions.
• Improve performance over time: By continuously updating probabilities, robots can learn from their experiences and improve their decision-making skills.
• Optimize task completion: Robots can plan and execute tasks with a higher degree of efficiency, even in uncertain or unpredictable conditions.

Robotic Path Planning Using Markov Models

Path Planning is the method of figuring out the best path a robot takes from its initial position to its end point. Designing an accurate model that predicts a point cloud is crucial for autonomous vehicles, drones, or warehouse bendable robots. The robot has to take account of many factors like an obstacle, consumption of energy and time taken etc.

How Markov Models Can Help Plan Optimal Robot Paths

Path planning can be described using Markov models, and the process is known as Markov Decision Processes (MDPs), in which radios are given for various paths or actions. It means that the robot is capable to assess its position, possible routes and the expected results of any performing phase.

So here is what it looks like in practice:

• The robot first defines a state space, representing all the possible locations and configurations.
• Then, it considers potential moves (i.e., driving straight; making a left turn; etc.).
• For any action the robot wants to perform—move left, avoid an obstacle, climb up a step—the robot weighs the relative successes and failures of that action, factoring in obstacles and terrain.
• The robot chooses the best path in such a way that highest of option probability at that time (goals), hits, it can be either reaching destination quickly or avoiding obstacles.

Markov Chain: A Simple Approach to Understanding Sequences

In terms of complexity, a Markov Chain is an MDP simplified to the degree that there are no available actions. It is a chain of connected events where each event has the probability conditional only to the previous state, not the entire history. Alternatively, a Markov Chain is the assumption that future events only rely on what state each element is currently in, regardless of how it got to that point.

Lets take an very example of robot moving on a grid. However, if the robot is in here and moves to there… then that transition only depends on where it currently is, not how it got there.

Markov Chain vs. Markov Decision Process

What differentiates a Markov Chain from a Markov Decision Process is the presence of choices. A Markov Chain is a series of probabilistic states where there are no choices involved while in a Markov Decision Process, we make the robot choose the best action to accomplish its aim.

Here’s a breakdown:

• Markov Chain: Focuses on transitions between states, with no actions taken.
• Markov Decision Process: Involves transitions between states where the robot actively selects actions that maximize rewards or minimize costs.

Markov Chains in Robotic Motion

When the robot has to move from one state to another, but no decisions need to be made, then Markov Chains are used for modeling the robotic motion. Some examples include:

• Motion Control: In simpler robots or devices, Markov Chains can describe a robot’s movement between fixed locations, such as turning on a conveyor belt or switching between different speeds.
• Sensor Data Interpretation: When robots receive sensor input (e.g., readings from cameras or LIDAR), Markov Chains can help predict the next likely sensor reading based on the current state of the robot.

Hidden Markov: Uncovering Hidden States in Robotics

The most important observation is that in many real-world situations, the true state of the robot will not actually be directly observable. The nature of this relationship leads to a Hidden Markov Model (HMM) because the robot state is “hidden” but observable outputs or indications exist, providing a window into that state.

For example, consider a robot navigating through a foggy environment. It can only sense limited data—such as distance to obstacles—but it cannot directly “see” the terrain or its exact position. In such cases, the robot uses HMMs to infer its state from the available sensory information, which can be noisy or incomplete.

How Hidden Markov Helps Robots in Unknown Environments

HMMs have the significant advantage as compare to visible or random stochastic model as they incorporate probabilities to make sense of incomplete or imperfect data. HMM (Hidden Markov Models) for Decision Making Using HMMs, robots can take decisions even when it does not have a complete information of its environment.

Here’s how robots use HMMs:

1. Sensor Data: The information supplied to the robot from any of its sensors (temperature, distance, sound etc.).
2. Hidden State Inference: Using these inputs, the probability that the robot is in a hidden state (say near some obstacles or in open space).
3. Decision Making: The robot uses this information to make decisions about where to move or what action to take next.

How Hidden Markov Helps in Robotics

• Robotic Navigation: Robots moving through unknown environments (e.g., dark rooms, maze-like areas) can use HMMs to infer where they are and how to best navigate.
• Speech Recognition: In robots with voice command functionality, HMMs allow them to “understand” and respond to spoken commands despite background noise or unclear speech patterns.
• Robotic Exploration: Robots exploring new terrains (e.g., search and rescue robots) use HMMs to infer hidden obstacles or hazards that aren’t immediately visible to their sensors.

Markov Decision: The Key to Smart Robot Behavior

The Markov Decision Process (MDP) lies at the core of many autonomous robotic systems. MDPs give the possibility to combine the likelihood nature of Markov Chains with decision making of robotics, which allow robots make decisions lead them to good outcomes with complex environment.

A long set of requirements has been addressed with paths that robots go through to balance those complicated factors towards competing dimensions along the axes of time and resources versus risk exposure to an uncertain environment, specifically Markov Decision Models. From a robot exploring a maze to another one cleaning the room to yet another doing complex assembly work, MDPs guide the robot on what action it should take at each step.

Wrapping Up: The Future of Robotics and Markov Models

Markov models transformed how robots decide, navigate through the environment and perform other complex tasks. MDPs and HMMs form the basis for a particle filter-based use of early MDPs that can cope with uncertainty, making real-time optimal decisions for robots.

Moving forward, these are pouring a lot of gas in the tank and making things much more capable for robots to do in an manner that requires less human assistance. And as robots get more sophisticated, probabilistic models will be key to making them work better than we ever could—as if they are a part of the same world.

The post Navigating Decisions in Robotics: An Introduction to Markov Models first appeared on Magnimind Academy.

What is the Random Forest?

What is Random Forest?

Random Forest is a versatile and powerful supervised machine learning algorithm. Basically, you can think of it as an ensemble model of decision trees. Random Forest can be used for both regression and classification problems in Python and R.

Before delving deeper into Random Forest, you need to understand the associated concepts first. So, here’s a snapshot of the related concepts.

Supervised machine learning

Machine learning is divided into three key categories namely supervised learning, unsupervised learning, and reinforced learning. In supervised machine learning, a training dataset is used to create the algorithm. Many different examples of inputs and outputs are used to train the algorithm so that it can learn how to classify fresh input data and predict future outcomes. 

Regression and classification

Both regression and classification are parts of supervised machine learning. We use regression algorithms to predict the output, which is a continuous or real value such as age, salary, price, etc. Classification algorithms are used to classify or predict the categorical output. For instance, an email spam filter is able to categorize every email into either of two classes – spam or not spam.

Decision trees

A decision tree is the most effective machine learning modeling technique widely used for classification and regression problems. When it comes to finding solutions, decision trees make hierarchical, sequential about the outcomes variable depending on the predictor data. There are three building blocks of a decision tree – a root node, branches, and leaves. To make it simple to understand, you can think of a decision tree as a flowchart, which draws an understandable pathway to an outcome or decision. It starts at one point and branches off into multiple directions. Each branch of a decision tree provides different outcomes possible. Basically, it’s a set of rules that we use to predict future data.

Basically, a Random Forest is only a bunch of decision trees grouped together. However, it’s important to understand that in a Random Forest all the trees are randomly mixed together. Therefore, when you use a single decision tree, it’ll come up with a set of rules based on the inputted training dataset. But when the same training dataset is inputted in a Random Forest algorithm, it’ll select features and observations randomly to build multiple decision trees and average the results of each of them.

How to create a Random Forest using decision trees?

How to create a Random Forest using decision trees?

To do this, you need to understand how to build a decision tree first. Since the key objective of a decision tree is to make the most favorable choice at each node’s end, it’s important to avoid impurity and reach maximum purity. To develop a better understanding of how a decision tree is built, you need to learn about its building blocks. 

Let’s try to understand the following fundamental concepts.

Entropy

We use entropy to measure the disorder, information, or purity of a dataset. If a dataset has mixed classes, it has more disorder and its entropy value will be higher. Therefore the potential for extracting information from it is also higher. On the contrary, a dataset with only one class doesn’t have much potential for information extraction.

Information Gain

We calculate the difference in entropy before and after we split the dataset to assess how ideal a feature is for splitting. Information gain refers to the amount by which a dataset’s entropy reduces due to the split. We use information gain when training decision trees as it helps to reduce uncertainty in them. A high information gain refers to the removal of a significant degree of uncertainty.

Gini Index

The Gini index is another measure for purity, which’s used by the Classification and Regression Tree (CART) algorithm. It is based on Gini impurity that measures how frequently a randomly selected element is incorrectly labeled in case it was labeled randomly according to distribution. If a dataset has two classes, the Gini index’s range remains between 0 and 0.5. In case the dataset is pure, it is 0, and if two classes are equally distributed it is 0.5. Therefore, we should try to have it reach 0 because that’s where it’ll be maximally pure and minimally impure.

Now that you’ve got an overview of the key mathematical concepts that are used in training decision trees, it’s time to take a look at Random Forest.

As we’ve already touched upon, the major difference between the Random Forest algorithm and the decision tree algorithm is that in the former, segregating nodes and establishing root nodes are randomly done. 

Bagging

In Random Forests, the usual technique of bagging is applied by the training algorithm to tree learners. In the bagging technique, different samples of training data are used instead of only one sample. Here, decision trees are trained on a subset of the actual training dataset. 

Bootstrapping of the training dataset

The subsets are obtained through random feature sampling and row sampling from the dataset – a method called bootstrapping. The sample datasets are reduced into summary statistics depending on the observation and combined by aggregation. We can Bootstrap Aggregation to reduce high variance algorithms’ variance.

Variance

Variance refers to an error that occurs from sensitivity to slight fluctuations in the training dataset. A high variance will lead to model noise or irrelevant data in the dataset rather than signal or the intended outputs – a problem called overfitting. Note that in training, an overfitted model can perform well but in an actual test, it won’t be able to differentiate the signal from the noise.

The majority rule

For the same input vector, slightly different predictions will be made by slightly differently trained trees. Generally, we apply the majority rule to decide on the final output. It means the prediction made by the majority of the trees is considered as the final output.

The key advantages and disadvantages of Random Forest

The key advantages and disadvantages of Random Forest

Random Forests offer a multitude of advantages – from relative ease of use to efficiency and accuracy. In addition, if you’re a data scientist and planning to utilize it in Python, you can use an efficient and simply random forest classifier library in scikit-learn. Let’s take a look at the pros and cons of Random Forests.

Advantages

• A Random Forest is significantly more efficient compared to a single decision tree when you’re performing analysis on a bigger database.
• Random Forests have the default ability to correct for the habit of overfitting of decision trees to their training datasets. You already know that overfitting results in inaccurate outcomes. You can almost entirely resolve the issue of overfitting when executing random forest algorithms with the help of random feature selection and bagging method.
• A neural network, which mimics the way of thinking of a human brain to disclose the underlying relationship present in a dataset, is more efficient than a Random Forest. However, a neural network is much more complicated than a Random Forest. Since it requires less expertise and time to build a Random Forest, the method often proves to be more useful for less experienced data science professionals.

Disadvantages

• Like any other tool, Random Forests also come with a few downsides. Since a Random Forest comprises several decision trees, it may require lots of memory, especially for larger projects. It may make it slower compared to some other algorithms.
• Decision trees are the building blocks of a Random Forest and decision trees frequently experience the issue of overfitting, which may affect the entire forest. However, generally, Random Forests prevent this problem by default. This is because they utilize features’ random subsets and use them to build smaller trees. The processing speed might get reduced due to it but accuracy will increase.

When you shouldn’t use Random Forests

There are some situations where Random Forest algorithms may not be the ideal option. These include the following.

• If you’re working with very sparse data, Random Forests may not generate good results. Here, an invariant space will be produced by the bootstrapped sample and the features’ subset, which may result in unproductive splits affecting the outcome.
• When it comes to performing extrapolation, random forest regression isn’t the best choice. This is the reason the majority of random forest applications are related to classification.

Parting thoughts

Once you’ve developed a good understanding of these fundamentals, you should join a machine learning course to master these concepts. These courses from a reputable institute will help you learn a wide range of other skills along with the Random Forest algorithm, even if you’re completely new to machine learning.

The post Creating A Forest From A Tree: A Brief Introduction To Random Forest first appeared on Magnimind Academy.

However, there’re lots of people who’re understandably concerned about the actual impact of artificial intelligence – whether it’s good for the society or not. Finding a proper answer to this question has become even more important as there’re individuals who’re deeply skeptical about AI’s potential and are wary of how the technology is going to impact the society.

Like any other technological aspect, the debate is complex for AI too. And of course, there’re benefits and issues along the way, as artificial intelligence becomes more pervasive in human lives.

We should always remember that when there’re challenges, there’re some good opportunities too. And we can’t afford to look at AI with skeptical eyes. Rather, we’ve to understand the benefits it has already offered to our society.

Major societal benefits of AI

Major societal benefits of AI

Some of the key societal benefits that artificial intelligence has brought along are outlined below.

Healthcare

Healthcare

Healthcare has always been one of the focal points of AI. It boasts of a huge amount of data to populate and analyze based on which computational sophistication has been improved by designers.

For instance, Merantix, a German company, applies deep learning to medical issues. It offers an application capable of detecting lymph nodes in the human body in CT (Computer Tomography) images. If the detection is done by humans, the charge would be prohibitively expensive. In this scenario, deep learning trains computers on datasets to learn what an irregular-appearing versus a normal-looking lymph node is. Once done, radiological imaging specialists apply this knowledge to real patients and identify the extent to which somebody is at risk of carcinogenic lymph nodes, at a significantly lower cost.

AI tools can predict substantial challenges lying ahead in advance and offer resources for patient education and proactive interventions, thus helping people to maintain their wellbeing.

Transportation

Transportation

Transportation is a field where artificial intelligence together with machine learning has produced major innovations. Autonomous vehicles like cars, buses, trucks etc use advanced capabilities that offer features like lane-changing systems, automated vehicle guidance, automated braking, use of sensors and cameras for collision avoidance, and analyzing information in real time by using AI, among others.

For instance, AI and LIDARs (light detection and ranging systems) play key roles in collision avoidance and navigation. These instruments provide information that helps to keep fast-moving vehicles in their designated lanes, thus helping them avoid other vehicles and applying brakes when needed etc, thus and ultimately saving human lives by reducing road accidents.

Prediction Of Natural Disasters

Prediction Of Natural Disasters

AI is considered as one of the perfect sources of predicting natural occurrences. There’s an AI model that can almost perfectly guide you what the weather will be for the next couple of days, which was almost unimaginable before the advent of artificial intelligence.

There’s also an incredible system that can predict, based on the simulation of tectonic plates of the earth, the time of volcanic eruptions. There’re AI-enabled projects that gather data for magnetometers of the phone and send it for analysis based on which successful predictions about earthquake can be made.

Farming

Farming

Farming is another sector that has been heavily benefitted from AI. This is an industry full of challenges like competition for natural resources, plateauing agricultural productivity, and rapidly growing population.

In this scenario, consider FarmLogs, a farming management app presently used by many farmers in the US, which uses technology and data to help farmers track the weather, monitor fields, obtain insights into soil utilizing historical satellite imagery, and even identify irregular plant growth. Real-time data analytics help farmers to maximize their crop yields and thus, in turn, their profits too.

Strengthened Economy

Strengthened Economy

Probably all of us have seen headlines that state something like adoption of AI will lead to unemployment. In reality, this is far from the truth. Artificial intelligence promotes a gradual evolution in the job field, which will be positive with the companies planning ahead.

Humans will still work, but they’ll be working more efficiently with the help of AI. Besides this unparalleled combination of machine and human, there will be a natural requirement of trained people who’ll be supervising the systems, apart from those who’ll actually do certain jobs. It’ll gradually result into more job openings, thus solidifying the economy.

Smart Cities

Smart Cities

AI is being implemented by various authorities to optimize different facilities. Artificial intelligence is considered as a way to deal with large volumes of data and to identify efficient ways of responding to various public requests. Instead of addressing service issues in ad hoc manners, authorities are implementing AI to be proactive in how urban services can be provided.

Smart city applications often use artificial intelligence to improve environmental planning, service delivery, energy utilization, resource management, and crime prevention, among others. Some of the top applications include intelligent traffic signals, e-governance applications, smart meters for utilities, Wi-Fi kiosks etc.

Overall Lifestyle

Overall Lifestyle

With the increasing implementation of AI in different segments of society, overall lifestyle of the humans gets enhanced. Some of the mundane tasks such as data entry or answering emails can be performed by intelligent assistants, freeing up precious time for humans to focus on creative aspects of the work.

Smart homes can be made capable of providing better security and reducing energy usage that would greatly promote the concept of a greener environment.

Steps we should take

With all the AI benefits, there comes some significant disadvantages as well, but that’s natural for any technology.

The question that is making lots of people worried is this: With too much of authority given to the machines, how can implementation of AI be made more favorable to the society? Or, at the least, how to make it not act like a threat to human life and property?

Transparency Is The Key

Transparency Is The Key

Results can be questioned even with a greatly planned decision-making system, if the reasoning can’t be demonstrated. For example, if AI has diagnosed an illness, the patient can always ask for a proper reasoning, failing which would lead to non-transparency of the system.

Assuming that artificial intelligence will be making hugely important decisions, implementation of AI has to be perfectly planned and the results have to be transparent and explainable to be accepted by the society. In addition, AI should be use data science to improve the living conditions.

Importance Of Ai Curriculum

Artificial intelligence should be used where it’s more effective and efficient to employ a machine to handle the task compared to engaging a human brain to perform it. So, companies should ensure scenarios where success gets replicated in raising the technology, in terms of both costs and resources.

High-quality protocols should also be developed by data scientists when it comes to selecting training data and taxonomies for AI. If it’s trained with patchy, skewed or flawed examples, the results will be going to be unreliable. So, it also needs to be ensured that the data is relevant, appropriate, accurate, diverse, accurately labeled, and representative.

For instance, if an AI-enabled system for hiring recommendations is trained on the data of present and past employees solely, and those employees aren’t diverse (e.g. predominantly older white females), the resulting model would likely be biased unfairly against candidates who’re young, racial minorities, and male. Even after an AI-enabled system is launched, the risk and quality of unfair bias need to be assessed by human overseers in ongoing new training data.

In addition, as we aim to improve the fairness and efficacy of AI with proper training data to benefit the society, we should also keep in mind that fundamental privacy principles are closely related to the massive amounts of data used by AI. Usage and retention of personal data needs to be minimized, while limiting the ways in which that data could be used in the future. Big data analytics solutions include the remedy of these problems.

Final takeaways

It’s important to note that AI isn’t able to learn on its own and thus, humans are required to help any type of artificial intelligence obtain a better understanding of all types of jobs, processes, things etc. When it comes to maximizing societal benefits, perhaps the best approach to leverage both AI systems and human-only systems is to do what each of them does best. Leveraging artificial intelligence as well as the best of human ability and values promises greater progress in accountability, transparency, and fairness. And this will be playing a crucial role in building a strong trust for AI in the society.

For instance, AI can be put to work to do the time-consuming analysis of the huge amount of information available. Building a culture of continuous learning and collaboration is crucial to take maximum advantage of artificial intelligence. This combined approach is what will make processes and people even more important than they are today. And to make the most out of the technology, society needs to deploy AI that puts humans first, protects human rights, and fosters humans’ trust. Immersive data science experience is proof that we can rely on artificial intelligence.

The post Invaluable Societal Benefits Of AI first appeared on Magnimind Academy.