Heuristic Search Techniques

• All the previous searches (BFS,DFS,DFID, and Bidirectional) have been blind searches
• They make no use of any knowledge of the problem
• If we know something about the problem, we can usually do much, much better
• Heuristic search techniques are Informed search techniques where we will have functions
to guide the search procedure.
• Heuristic is a Rule of Thumb that use domain specific knowledge to estimate the quality or
potential of partial solutions.
• Heuristics technique is a criteria for deciding which among several alternatives will be the
most effective in order to achieve some goal.
• Heuristic is a technique that no longer guarantees to find the best answer but almost always
finds a very good answer.
Heuristic Search Techniques (Contd.)
• Using good heuristics, we can hope to get good solution to hard problems in less than
exponential time.
• There is no general theory for finding heuristics, because every problem is different
• There are general-purpose heuristics that are useful in a wide variety of problem
domains.
• We can also construct special purpose heuristics, which are domain specific.
Heuristic Search Techniques (Contd.)
• The searches which use some domain knowledge are called informed search strategies.
• Formally, informed search problem is a 5-tuple [S, S0, O, G, h].
• h( ) is a heuristic function estimating the distance to a goal.
• The function h(n) is the estimated cost of the cheapest path from node n to a goal node.
Heuristic Search Techniques (Contd.)
Branch & Bound Search
• Sometimes, it is called Uniform Cost Search.
• In Branch & Bound we will introduce the notion of cost.
• While generating a search space, it expands the least-cost partial path.
• The cost function g(X) or C(X) assigns cumulative expense to the path from Start node to
the current node X by applying the sequence of operators .
• If operators have unit cost, then this is same as BFS.
Heuristic Search Techniques (Contd.)
Branch & Bound Algorithm
Heuristic Search Techniques (Contd.) Contents of OPEN list are shown below:

Branch & Bound Algorithm (contd.)

Let us consider an example state space
and apply B&B algorithm:

Contents of CLOSED list are shown below:

Heuristic Search Techniques (Contd.)
Hill Climbing
• Hill Climbing is an optimization technique that belongs to the family of local searches.
• Local search algorithms operate using a single current node and generally move only to
neighbors of that node.
• It is a local search algorithm that continuously moves in the direction of increasing value –
that is uphill to find the best solution to the problem.
• It terminates when it reaches a “peak” where no neighbor has a higher value.
• The algorithm does not maintain a search tree as it only keeps a single current state.
• It is also called greedy local search as it only looks to its good immediate neighbor state and
not beyond that.
• Moving through a tree of paths, hill climbing proceeds in depth-first order, but the choices
are ordered according to some heuristic value.
Heuristic Search Techniques (Contd.)
Hill Climbing Algorithm
Input: Start and goal states of the problem
Output: Yes or No
Method:
1. Initialize: Set OPEN = {S0} ; current-state = start state;
2. Fail: If OPEN ={ }, terminate with failure
3. Select: Select the current state, n, from OPEN
4. Terminate: If n G (if n is a goal state), terminate with success.
5. Expand:
Generate the successors of n using state transition operators (O).
For each successor m, compute h(m). (i.e., estimated cost)
If a successor m is better than the current state n, then make the state m as the current
state and insert it into OPEN.
6. Loop: Go to step 2.
Heuristic Search Techniques (Contd.)
Problems in Hill Climbing
• The search process may reach to a position that is not a solution but from there no move
improves the situation.
• This will happen if we have reached a Local maximum, a plateau or a ridge.
• Local maximum: It is a state that is better than all its neighbors but is not better than
some other states farther away. From this state all moves appear to be worse. Since hill-
climbing uses a greedy approach, it will not move to the worse state and terminate itself.
The process will end even though a better solution may exist.
• Solution to this is to backtrack to some earlier state and try going in different
direction.
Heuristic Search Techniques (Contd.)
Problems in Hill Climbing (Contd.)
• Plateau: It is a flat area of the search space in which, a whole set of neighboring states
have the same value. It is not possible to determine the best direction.
• Here make a big jump to some direction and try to get to new section of the search
space.
• Ridge: It is an area of search space that is higher than surrounding areas, but that can not
be traversed by single moves in any one direction. (Special kind of local maxima).
• Here apply two or more rules before doing the test i.e., moving in several directions at
once.
Heuristic Search Techniques (Contd.)
Hill Climbing Algorithm (Contd.)
Consider the Blocks World Problem:

The possible operators for this problem are as follows:

■ Pick up one block and put it on the table.

■Pick up one block and put it on another one.

Suppose we use the following heuristic function:

■ A Local heuristic function: Add one point for a block if it is in correct location and
subtract one point if it is in wrong location. (Count +1 for every block that sits on the
correct thing. Count -1 for every block that sits on an incorrect thing. )
Heuristic Search Techniques (Contd.)
Hill Climbing Algorithm (Contd.)
Heuristic Search Techniques (Contd.)

Hill Climbing Algorithm (Contd.)

Now let us consider another heuristic function.
A Global heuristic function: Add one point for every block in a correct structure that the
block is sitting on. Subtract one point for every block in a wrong structure (For each block
that has the correct support structure: +1 to every block in the support structure. For each
block that has a wrong support structure: -1 to every block in the support structure.).
■ Using this function, the goal state has score 11. The initial state has the score -1.
Heuristic Search Techniques (Contd.)
Hill Climbing Algorithm (Contd.)
Heuristic Search Techniques (Contd.)
Beam Search
• Beam Search progresses level by level.
• It moves downward from the best W nodes only at each level. Other nodes are ignored.
• W is called width of beam search.
• It is like a BFS where also expansion is level wise.
• Best nodes are decided on the heuristic cost associated with the node.
• If B is the branching factor, then there will be only W*B nodes under consideration at any
depth but only W nodes will be selected.
• If W=1, then it becomes hill climbing search where always best node is chosen from
successor nodes.
Heuristic Search Techniques (Contd.)
Algorithm – Beam search
Heuristic Search Techniques (Contd.)
Beam search (Contd.)
The search tree generated using beam search algorithm , assume W=2 and B=3 is given
below.
Heuristic Search Techniques (Contd.)
Best-First Search
• Expand the best partial path from current node to the goal node.
• The cost of the partial path is calculated using some heuristic.
• The name Best-first refers to the method of exploring the node with the best “score” first.
• If the state has been generated earlier and new path is better than the previous one, then
change the parent and update the cost.
• It should be noted that in hill climbing, sorting is done on the successors nodes, where as
in the Best-first search, sorting is done on the entire list.
• It is not guaranteed to find an optimal solution, but generally it finds some solution faster
than solution obtained from any other method.
• The performance varies directly with the accuracy of the heuristic evaluation function.
Heuristic Search Techniques (Contd.)

Best First Search Algorithm

Input: Start and goal states of the problem
Output: Yes or No
Method:
1. Initialize: Set OPEN={S0}; CLOSED={ }
2. Fail: If OPEN={ }, terminate with failure
3. Select: Select the best score node, n, from OPEN and save it in CLOSED.
4. Terminate: If n G (if n is a goal state), terminate with success.
5. Expand: Generate all successors of node n using state transition operators (O).
• Assign the successor nodes a score using the evaluation function and add the
scored nodes to OPEN.
• Sort the list OPEN according to their heuristic values.
6. Loop: Go to step 2
Heuristic Search
A* Algorithm
Techniques (Contd.)
• A* (“Aystar”) (Hart, 1972) method is a combination of ‘Branch and Bound’ and ‘Best-
first search’, combined with the dynamic programming principle.
• It uses a heuristic or evaluation function usually denoted by f(n) to determine the order in
which the search visits nodes in the tree.
• The heuristic function (or Evaluation function) for a node n is defined as f(n) = g(n) + h(n)
• The function g is a measure of the cost of getting from the Start node (initial state) to the
current node n.
• The function h is an estimate of additional cost of getting from the current node n to the
Goal node (final state).
• Starting with a given node, the algorithm expands the node with the lowest f(n) value.
Heuristic Search Techniques (Contd.)
The A* formula: f(n) = g(n) + h(n)
start
• g(n) is the (known) distance from start to n
• h(n) is the (estimated) distance from n to a
goal
g(N)
• f(n) is just the sum of these
• f(n) is our best guess as to the distance from
f(N) start to a goal, passing through n.

h(N)

goal
Heuristic Search Techniques (Contd.)
A* Algorithm
Heuristic Search Techniques (Contd.)
A* Algorithm (Contd.)
Let us consider an example state space Contents of OPEN list are shown below:
and apply A* algorithm:

Contents of CLOSED list are shown below:

Heuristic Search Techniques (Contd.)
Example – Eight Puzzle Problem
Solve Eight puzzle problem using A* algorithm
Start state Goal state

3 7 6 5 3 6

5 1 2 7 2

4 8 4 1 8

• Evaluation function f(n) = g(n) + h(n)

h(n) = the number of tiles not in their goal position in a given state n
g(n) = depth of node n in the search tree
• Initial node has f(initial_node) = 4
• Apply A* algorithm to solve it.
• The choice of evaluation function critically determines search results.
Heuristic Search Techniques (Contd.)
Example – Eight Puzzle Problem (Contd.)
Start State
Search Tree f = 0+4
3 7 6
5 1 2
4 8

up left right
(1+3) (1+5) (1+5)
3 7 6 3 7 6 3 7 6
5 2 5 1 2 5 1 2
4 1 8 4 8 4 8

up left right
(2+3) (2+3) (2+4)
3 6 3 7 6 3 7 6
5 7 2 5 2 5 2
4 1 8 4 1 8 4 1 8

left right
(3+2) (3+4)
3 6 3 6
5 7 2 5 7 2
4 1 8 4 1 8
down
(4+1)
5 3 6 right 5 3 6 Goal
7 2 7 2 State
4 1 8 4 1 8
Heuristic Search Techniques (Contd.)
Harder Problem
• The quality of solution will depend on heuristic function.
• Harder problems (8 puzzle) can’t be solved by heuristic function defined earlier.

• A better estimate of h function might be as follows. The function g may remain same.
h(n) = the sum of the distances of the tiles from their goal position in a given state n.
■ This is called Manhattan distance heuristic.
• Initial node has h(initial_node) = 1+1+2+2+1+3+0+2=12
Heuristic Search Techniques (Contd.)
Optimal solution by A* Algorithm
A* algorithm finds optimal solution if heuristic function is carefully designed and is
underestimated.
Underestimation
• If we can guarantee that h never overestimates actual value from current to goal, then A*
algorithm is guaranteed to find an optimal path to a goal, if one exists.
Heuristic Search Techniques (Contd.)
Explanation -Example of Underestimation

Assume the cost of all arcs to be 1. A is expanded to B, C and D.

‘f’ values for each node is computed.
B is chosen to be expanded to E.
We notice that f(E) = f(C) = 5
Suppose we resolve in favor of E, the path currently we are expanding. E is expanded to F.
Clearly expansion of a node F is stopped as f(F)=6 and so we will now expand C.
Thus we see that by underestimating h(B), we have wasted some effort but eventually
discovered that B was farther away than we thought.
Then we go back and try another path, and will find optimal path.
Heuristic Search Techniques (Contd.)
Example – Overestimation
Here h is overestimated • A is expanded to B, C and D.
• Now B is expanded to E, E to F and F to G
for a solution path of length 4.
• Consider a scenario when there a direct
path from D to G with a solution giving a
path of length 2.
• We will never find it because of
overestimating h(D).
• Thus, we may find some other worse
solution without ever expanding D.
• So by overestimating h, we can not be
guaranteed to find the cheaper path
solution.
Heuristic Search Techniques (Contd.)
Admissibility of A*
• A search algorithm is admissible, if for any graph, it always terminates in an optimal path
from initial state to goal state, if path exists.
• If heuristic function h is underestimate of actual value from current state to goal state,
then it is called admissible function (h(n)≤h*(n)).

• Alternatively we can say that A* always terminates with the optimal path in case
• h(n) is an admissible heuristic function.
Heuristic Search Techniques (Contd.)
Monotonic Function
• A function h is monotone if
•  states Xi and Xj such that Xj is successor of Xi
h(Xi) – h(Xj) ≤ cost (Xi, Xj)
where, cost (Xi, Xj) actual cost of going from Xi to Xj
• h (goal) = 0
• In this case, heuristic is locally admissible i.e., consistently finds the minimal path to each
state they encounter in the search.
• With monotonic heuristic, if a state is rediscovered, it is not necessary to check whether the
new path is shorter.
• Each monotonic heuristic is admissible
• A cost function f(n) is monotone if f(n) ≤ f(succ(n)), n.
• For any admissible cost function f, we can construct a monotonic admissible function.
Heuristic Search Techniques (Contd.)
Iterative Deepening A* (IDA*) Algorithm
• Iterative deepening A* (IDA*) is a combination of the Depth-First Iterative Deepening
(DFID) and A* algorithm.
• But, instead of using depth-bound we use f-limit.
• At each iteration, perform a DFS pruning off a branch when its total cost (g+h) exceeds a
given threshold.
• The initial threshold starts at the estimate of the cost of the initial state, and increases for
each iteration of the algorithm.
• At each iteration, the threshold used for the next iteration is the minimum cost of all values
exceeded the current threshold.
• These steps are repeated till we find a goal state.
Heuristic Search Techniques (Contd.)

Iterative Deepening A* (IDA*) Algorithm

Algorithm works as follows.
• Start with f-limit = h(start) and we do a DFS.
• Prune any node if f(node) > f-limit.
• Next f-limit = min-cost of any node pruned.
• These steps are repeated till we find a goal state.
Ist iteration ( Threshhold = 5) O 5

O O O O O
6 8 4 8 9
   

IInd iteration ( Threshhold = 6) O 5

O O

O O O O O O
7 5 9 8 4 7
   

IIIrd iteration ( Threshhold = 7) O 5

O O

O O O

O O O O O O Goal
8 9 4 4 8
  

Iterative Deepening A* Example

Heuristic Search Techniques (Contd.)
Contd..
• Given an admissible monotone cost function, IDA* will find a solution of least cost or
optimal solution if one exists.
• IDA* not only finds cheapest path to a solution but uses far less space than A* and it
expands approximately the same number of nodes as A* in a tree search.
• An additional benefit of IDA* over A* is that it is simpler to implement, as there are no
open and closed lists to be maintained.
• A simple recursion performs DFS inside an outer loop to handle iterations.