 Hello everyone, this is Alice Gao. In the previous video, I introduced the story on a male pickup robot. And I discussed how to define different nodes or different variables for the story. So we came up with four different nodes, one for accident, which is a random variable, two decision nodes, one for whether to choose the short route or the long route, and the other one for wearing pads or not. And then there's also a utility node, which is denoting the robot's happiness in each state. In this video, we are going to keep working on constructing the decision network. So now that we have all the nodes, we are going to try to connect the nodes together. In order to connect the nodes together, we have to figure out how do the random variables and the decision variables influence each other, right? Does a random variable come first, and then we observe the outcome? Or do we have to make a decision first before we observe the outcome of a random variable? So for this question here, recall that we came up with four different nodes. This is on the previous slide, accident, short, pads, and utility. So let's look at the chance nodes and the decision nodes. So accident, short, and pads. And let's think about how do these nodes affect each other? Think about this yourself for a second, and then keep watching for the answer. Here are the answers. So first of all, what is a right order of thinking about these three variables? Well, in the story, the robot has to make both decisions before it's going to observe whether there's an accident or not, right? Actually, depending on which route the robot takes, there may or may not be a positive chance of having an accident. And also, the robot has to decide on whether to wear pads or not before it will observe whether there's an accident. So in terms of time, the decision nodes come first, right? And then the accident node comes later, okay? So given this, there might be arrows pointing from the decision nodes to the random variable accident. So let's think about, is there a influence or not? So whether the robot chooses the short route or not does affect the probability of an accident, right? In fact, if the robot chooses the long route, an accident will not occur because the long route doesn't have that problem. If the robot chooses the short route, then there is a positive chance that an accident may occur. So short definitely affects accident. On the other hand, wearing pads or not does not affect accident at all, right? Because as the story mentioned, wearing pads or not does not affect the probability that an accident happens. It only affects the severity of the damage if an accident actually happens. Given our analysis here, let's connect the nodes based on this. I created an edge from short to accident because short affects accident. There's no edge from pass to accident because wearing pads or not does not influence the probability that an accident occurs. Now similar to how we have a conditional probability distribution attached to every node in the Bayesian network, we will also have one such distribution attached to every chance node in the network. But in this case, notice that accident gets affected by a decision node. So it doesn't get affected by another chance node, but that's okay. We can still come up with a conditional probability distribution. So in this case, we have two cases, the probability of accident given that we go on the short route. And the probability of accident given that we go on the long route. Now for the long route, it's easy because an accident will never occur on the long route, the probability is zero. For the short route, there is a fixed probability for an accident to occur. And we don't know what this number is. In fact, later on, we are going to think about if we change this probability, then how does our decision change? So for now, let's set this to be a variable. So let's say this is denoted by a variable Q, which is a number between zero and one, right? So these two lines make up the conditional probability distribution for the node accident. I'm going to stop here for this video. In the next video, we will think about the utility node. So which edges will go into the utility node and how should we specify the utility function for the robot? Thank you very much for watching. I will see you in the next video. Bye for now.