 I am Mr. Ninath Kulkarni working as an assistant professor in department of mechanical engineering at Vulture Institute of Technology, Solapur. In today's session, I will be dealing with how to solve some simple problems based on AI that is artificial intelligence. At the end of the session, students will be able to solve some simple problems using AI technique by the production system method. Now, what is problem solving? Problem solving is solving the problem. We are given a certain problem. We are given the initial description of the problem and we have to reach the solution. We have to reach the desired solution by solving some simple steps by going through some simple steps just like the proofs, mathematical proofs, the theorem proofs such as Pythagoras theorem. If we are going to prove the Pythagoras theorem, we have to solve some steps. We have to go through the some steps so that we reach the desired solution that is the hypotenuse square is equal to side 1 square plus side 2 square. This solution is the desired solution. So, we have to go through some steps. A problem can be considered to consist of a goal. So, there is a goal and we have to do some certain set of actions that can be taken so that the goal is reached. At any given time, we consider the state of search space to represent where we have reached. So, at any instance of time, we have to specify the state of the search space. What is the status of our solution? To define a problem as a search space, some preliminary concepts are there. First concept is state. State indicates the status of a solution at a given step of the problem solving procedure. So, at any given step, what is the status of your solution is there? That is called as a state. Initial state is the initial status of the solution that is the initial condition of the solution. Goal state is the desired solution whatever we require. An operator is that constituent that transforms one state to another state in order to reach closer towards your goal state that is the desired solution. So, the operator it is that part which gives the motion to the solution and what is mean by state search or state space approach is that the process of applying an operator that is a force to state and its subsequent transition from one state to next state and this process is continued until we reach a goal state. So, until the goal state is reached, this process is continued or iterated that process is called as the this approach is called as the state space search. For example, while playing chess we know that initial state is the starting position or starting position on the board. The goal state is the position in which at least one player reaches the winning position that means one player checkmates the another player that is the winning state and the algorithm continues until the goal state or winning position is reached. This is how the state space search approach is worked. In AI, for formally describing a problem we have to define first all the state space that means it contains all the possible configurations of the relevant objects that means it consists of all the status of solutions. After forming or after defining a state space we have to define some of these states as possible initial states. Also, we have to define some goal state there may be a single goal state or in multiple goal states. So, in games such as chess or tic-tac-toe there may be multiple goal states. Next we have to specify a certain set of rules as the possible actions that are allowed if this happens then this should be done. So, these rules are stored in the knowledge space. There are two types of learning as in the search based algorithms. One is the monotonic learning in which the agent may not learn any knowledge that contradicts what it already knows. That means the agent, the search agent it does not update itself, it does not update the rules and it follows what is there in the algorithm. Second one is non-monotonic learning in which the agent learns new knowledge that contradicts the existing knowledge. So, in non-monotonic the algorithm is updated the knowledge base is updated. This is the algorithm of problem solving. First step is analyze the problem to get the starting state and goal state. The first step is we have to specify the starting state that is initial state and goal state. After that find out the data about starting state and goal state. Third step is find out the production rules from initial database for proceeding the problem to goal state. So, we have to specify the production rules or control strategy. Next we have to select some of these rules, some best rules from the set of rules that can be applied to this problem. After that we have to apply these rules to initial state that means the first state and we have to proceed to get to the next state until we reach the goal state. Determine some new generated states and make them as the current state. We call them as the intermediate state. And finally we have to achieve some information and reach towards the goal state. Whenever we reach the goal state we have to exit. This is the general algorithm for problem solving and in problem solving one model or one computational model is there that is production system. So, this production system is a computational model that can be applied to implement search algorithms and model human problem solving. In this the knowledge is packed in the form of little quanta these are called as productions. So, production is a rule consisting of situation, recognition part and action part that means if a situation is given what action one should take. This if then is coded in the production if A then B. A production is a situation action pair in which left side is a list of things to watch for and right side is the action to do. So, production system consists of a set of production rules which are in the form of A tends to B if A then B. Second is database which consists of all the appropriate information of the particular task or problem. Next a control strategy how the rules are prioritized and if several rules are active simultaneously which rule should come first that conflicts are resolved. Next a rule applyer means it is a software which checks whether this rule should be applied to this situation or not. So, these are the four components of production system. These are advantages or features of production system first is simplicity because if if then rules are there that is why it is simple. Second modularity because there are discrete modules of rules and they can be easily separated. The modifiability is the capacity to modify the rules existing rules and this production system is knowledge intensive, the knowledge base it only stores all the information about the problem. But there are certain limitations first is the opacity if the prioritization of rules it is not proper then it may lead to the algorithm being opaque. During execution of the program several rules may be active which makes this algorithm inefficient. This algorithm does not store the result for future use that means if there is an extension of one problem that can be applied to another then it starts from the first and when a new rule is added to a database it should ensure that it does not conflict with the existing rules that conflict resolution should be there. Now I will take you through the water jug problem. Suppose you are given three jugs A, B and C, A has capacity of 8 liters, B has 5 liters and C has 3 liters, but they are not calibrated. Jug A initially it is filled with 8 liters and B and C are empty and by series of pouring back and forth among three jugs we have divide the 8 liters into two equal parts that is 4 liters in jug A and 4 liters in jug B. So this is the problem and we have to solve this. The initial condition is that A is filled with 8 liters and B and C are empty that means initial state are A 8 0 0 and your goal state is 4 4 and 0 this is the goal state. Now we are going to proceed with this problem. The initial state is 8 0 0 and from this initial state we have three options. First option is we have going to put 3 liters in jug C that is 3 liter capacity. So the state will come as 5 0 3. Second option is 3 5 0 that means we are going to pour 5 liters in jug B and third option is we are going to make the jug A empty and 5 and 3 liters are poured in jug B and C. These are the three options and from these three options we will follow this hierarchy and these are the solutions that we will get. So this algorithm will search each and every state of the solution and until we reach this end state. So this is the end state. So whenever this algorithm reaches end state it stops and exits. So this is the strategy we follow in the production systems algorithm until we reach the end state. So these are the references that are referred and thank you.