A set of five levels is used to train the model. As NMF is a mathematical model, it doesn’t accept characters directly. Our game description is in characters form. So, we make game in a form so that mathematical operations are performed, and it accepts our character input.

For this purpose character to binary conversion is performed.We first convert character matrix W into binary matrix. We have 8 parameters w, 1, 2, 3,., A, g, +. In Matrix where “w “is present, it means that wall is generated. In Matrix where “1 “is present, it means that monster is generated, and it has a quick moment. In Matrix where “2” is present, it means that monster is generated, and it has a normal moment. In matrix where “3” is present, it means that monster is generated, and it has a slow moment. In matrix where “.” is present it means that floor generates. In matrix where “A” is present, it means that key is not generated but player creates. In matrix where “g” is present, it means that it is the starting point of player and it is player/’s goal. In Matrix where “+ is present it means that After key collection, the player achieves a goal and level changes.

When we tune our parameters, it gives us different results. Earlier when we tune it gives us a zeros matrix, so we tune these parameters to get a binary output.

If the element is present in the matrix then it is added as “1” otherwise “0”.

First parameter is w. w parameter is of 9×13.In “w” parameter we have 9 rows and 13 columns. In matrix where w is present, it is represented as “1” otherwise” 0″.And we have 5 levels. It means that we have 45×13 matrix of w. This is training set of NMF\_w matrix.

For 1 parameter we have 9×13 matrix. It means that we have 9 rows and 13 columns, so for 1 parameter as levels are 5, so we have 45×13 matrix. In matrix where 1 is present, it is represented as “1” otherwise “0”. This is training set of NMF\_1matrix.

For “2” parameter we have 9×13 matrix. It means that we have 9 rows and 13 columns, so for 2 parameter as levels are 5, so we have 45×13 matrix. In a matrix where 2 is present, it is represented as “1” otherwise “0”. This is training set of NMF\_2matrix.

For parameter “3” we have 9×13 matrix. It means that we have 9 rows and 13 columns, so for 3 parameter as levels are 5 so we have 45×13 matrix. In matrix where 3 is present, it is represented as “1” otherwise “0”. This is training set of NMF\_3 matrix.

For parameter “.” we have 9×13 matrix. It means that we have 9 rows and 13 columns, so for. parameter as levels are 5 so we have 45×13 matrix. In matrix where. is present it is represented as “1” otherwise “0”. This is training set of NMF\_. matrix.

For parameter “A” we have 9×13 matrix. It means that we have 9 rows and 13 columns, so for A parameter as levels are 5, so we have 45×13 matrix. In matrix where A is present, it is represented as “1” otherwise “0”. This is training set of NMF\_A matrix.

For parameter “g” we have 9×13 matrix. It means that we have 9 rows and 13 columns, so for g parameter as levels are 5 so we have 45×13 matrix. In matrix where g is present, it is represented as “1” otherwise “0”. This is training set of NMF\_w matrix.

For parameter “+” we have 9×13 matrix. It means that we have 9 rows and 13 columns so for the + parameter as levels are 5 so we have 45×13 matrix. In matrix where + is present, it is represented as “1” otherwise “0”. This is training set of NMF\_+ matrix.

Now the binary matrix generation.For w parameter,the training set matrix of 45×13 and a input prediction matrix of 9×13 and get the new matrix of w.

For 1 parameter,the training set matrix of 45×13 and a input prediction matrix of 9×13 and get the new matrix of w.

For 2 parameter,the training set matrix of 45×13 and a input prediction matrix of 9×13 and get the new matrix of w.

For 3 parameter,the training set matrix of 45×13 and a input prediction matrix of 9×13 and get the new matrix of w.

For . parameter,the training set matrix of 45×13 and a input prediction matrix of 9×13 and get the new matrix of w.

For A parameter,the training set matrix of 45×13 and a input prediction matrix of 9×13 and get the new matrix of w.

For g parameter,the training set matrix of 45×13 and a input prediction matrix of 9×13 and get the new matrix of w.

For + parameter,the training set matrix of 45×13 and a input prediction matrix of 9×13 and get the new matrix of w.

To estimate an NMF Model, the multiplicative update algorithm is used which is shown below: