Artificial Intelligence for Decision Making

Question

Do you know any example for this topic?. I have searched google but had no luck with any Decision Making using Artificial Intelligence example ( at least any truly developed)

Answer 1

i have an example for decision making in AI. it is about choosing a dress to purchase. It asks for inputs as follows: 1) Whether you are Male or Female 2)Price range must be less than 1000 or greater than 1000 3)Age <20 or >20

Based on these inputs , my code will analyse which type of dress suits you and in which shop you can purchase it. I have just given few knowledge in my code. Only 3 shop's names i have kept in my knowledge base. It gives output as a percentage . Eg) Shop A: 20.5% Shop B: 40.5& Shop C: 39.0% This means that you have high probability of finding your favourite dresses in Shop B.

    $   import java.util.*;
import java.lang.*;
import javax.swing.*;
import java.util.Collection.*;
import java.util.Set.*;
public class dress1
{
    static int i,j;

    static float[] count=new float[3];static int[] count1=new int[3];
    static String[] shop=new String[] {"Jean and Tops","Jean and T-shirt","Jean and Shirt/T-shirt","Sherwani","Pyjamas","Formal shirt and Pant","Shorts and shirt","Coat Suit","Sari","Skirt and Tops","Ghagra Choli","Salwar kameez","Churidaar","Pattu Pavadai","Frock"};
    static int[] divac= {10,9,8,1,1,2,1,1,7,1,1,1,5,1,1};
    static int[] kalac=  {1,1,1,2,1,1,1,1,10,1,8,2,2,9,1};
    static int[] megac= {7,5,6,3,2,8,2,6,2,4,2,1,2,2,2};
    static String[] selected=new String[15];
    static boolean value;
    public static void main(String arg[])
    {

    count[0]=0;count[1]=0;count[2]=0;
    Set uni = new HashSet();
    uni.add("Jean and Tops");
    uni.add("Jean and T-shirt");
    uni.add("Sari");
    uni.add("Skirt and Tops");
    uni.add("Ghagra Choli");
    uni.add("Salwar kameez");
    uni.add("Churidaar");
    uni.add("Pattu Pavadai");
    uni.add("Sherwani");
    uni.add("Frock");
    uni.add("Formal shirt and Pant");
    uni.add("Jean and Shirt/T-shirt");
    uni.add("Shorts and shirt");
    uni.add("Coat Suit");
    uni.add("Pyjamas");

    Set male = new HashSet();
    male.add("Sherwani");
    male.add("Pyjamas");
    male.add("Formal shirt and Pant");
    male.add("Jean and Shirt/T-shirt");
    male.add("Shorts and shirt");
    male.add("Coat Suit");
    Set fem = new HashSet();
    fem.add("Jean and Tops");
    fem.add("Jean and T-shirt");
    fem.add("Sari");
    fem.add("Skirt and Tops");
    fem.add("Ghagra Choli");
    fem.add("Salwar kameez");
    fem.add("Churidaar");
    fem.add("Pattu Pavadai");
    fem.add("Frock");

    Set l20=new HashSet();
    l20.add("Jean and Tops");
    l20.add("Jean and T-shirt");
    l20.add("Skirt and Tops");
    l20.add("Churidaar");
    l20.add("Pattu Pavadai");
    l20.add("Frock");
    Set g20=new HashSet();
    g20.add("Jean and T-shirt");
    g20.add("Sari");
    g20.add("Salwar kameez");
    g20.add("Churidaar");

    Set ml20=new HashSet();
    ml20.add("Pyjamas");
    ml20.add("Shorts and shirt");
    Set mg20=new HashSet();
    mg20.add("Sherwani");
    mg20.add("Formal shirt and Pant");
    mg20.add("Coat Suit");
    mg20.add("Jean and Shirt/T-shirt");

    Set mpl1000= new HashSet();
    mpl1000.add("Pyjamas");
    mpl1000.add("Shorts and shirt");
    mpl1000.add("Formal shirt and Pant");
    Set mpg1000=new HashSet();
    mpg1000.add("Sherwani");
    mpg1000.add("Coat Suit");
    mpg1000.add("Jean and Shirt/T-shirt");

    Set pl1000=new HashSet();
    pl1000.add("Frock");
    pl1000.add("Skirt and Tops");
    pl1000.add("Churidaar");
    pl1000.add("Salwar kameez");
    Set pg1000=new HashSet();
    pg1000.add("Jean and Tops");    
    pg1000.add("Jean and T-shirt");
    pg1000.add("Sari");
    pg1000.add("Ghagra Choli");
    pg1000.add("Pattu Pavadai");

    Scanner input=new Scanner(System.in);
    System.out.println("Enter M for Male and F for Female");
    String st=input.nextLine();
            Set int1 = new HashSet (uni);
    if(st.equals("M"))
    {
        System.out.println("Male");
        int1.retainAll(male);

    Scanner input1=new Scanner(System.in);
    System.out.println("Enter 1 if age <20 or enter 2 if age > 20");
    String st1=input1.nextLine();
            Set int2= new HashSet (int1);
    if(st1.equals("1"))
           {
    System.out.println("Age Less than 20 Male");
    int2.retainAll(ml20);
    }
    else if(st1.equals("2"))
    {
        System.out.println("Age Greater than 20 Male");
        int2.retainAll(mg20);
    }
    Scanner input2=new Scanner(System.in);
    System.out.println("Enter 1.Price Range < 1000 or 2. Price Range >1000");
    String st2=input2.nextLine();
            Set int3= new HashSet (int2);
            if(st2.equals("1"))
            {
                System.out.println("Price Less than 1000 ,Male");
                int3.retainAll(mpl1000);
            }
            else if(st2.equals("2"))
            {
                System.out.println("Price Greater than 1000 ,Male");
                int3.retainAll(mpg1000);
            }

            i=0;    
            for(Object object : int3)
            {
                String element = (String) object;
                selected[i++]=element;
                System.out.println(element);
            }
            calc();

        } /*end of male*/

        else if(st.equals("F"))
        {

            System.out.println("Female");
                int1.retainAll(fem);
            Scanner input1=new Scanner(System.in);

            System.out.println("Enter 1 if age <20 or enter 2 if age > 20");
            String st1=input1.nextLine();
            Set int2= new HashSet (int1);
            if(st1.equals("1"))
            {
                System.out.println("Age Less than 20 Female");
                int2.retainAll(l20);
            }
            else if(st1.equals("2"))
            {
                System.out.println("Age Greater than 20 Female");
                int2.retainAll(g20);
            }   
    Scanner input2=new Scanner(System.in);
    System.out.println("Enter 1.Price Range < 1000 or 2. Price Range >1000");
    String st2=input2.nextLine();
            Set int3= new HashSet (int2);
            if(st2.equals("1"))
            {
                System.out.println("Price Less than 1000 ,Female");
                int3.retainAll(pl1000);
            }
            else if(st2.equals("2"))
            {
                System.out.println("Price Greater than 1000 ,Female");
                int3.retainAll(pg1000);
            }

            i=0;
            for(Object object : int3)
            {
                String element = (String) object;
                selected[i++]=element;
                System.out.println(element);
            }
            calc();
        }/*end of female*/

    }/*end of main*/

    public static void calc()
    {
        float finalcount=0;


        int k=0;

        while(k<i)
        {
            for(j=0;j<15;j++)
            {
                value=selected[k].equals(shop[j]);
                if(value)
                {
                    break;
                }
            }/*end of j*/
            count[0]=count[0]+divac[j];
            count[1]=count[1]+kalac[j];
            count[2]=count[2]+megac[j];
            k++;
            }/*end of while*/

            for(int c=0;c<3;c++)
            {
            finalcount=finalcount+count[c];
            }
            for(int c=0;c<3;c++)
            {

                count[c]=((count[c]*100)/finalcount);
            }
            System.out.println("Availability of dresses available in various shops are:");
            System.out.println("Diva is "+count[0]+"%");
            System.out.println("Kalanikethan is "+count[1]+"%");
            System.out.println("Megamart is "+count[2]+"%");
    }
 }/*end of class*/

Answer 2

Decision-making (DM) is a topic widely studied by different fields of science. (Cognitive sciences, Neurosciences, Computer sciences, Behavioral Economics, Operations research, etc. *1)

However, DM problems are varied and the computational approach to address that problem will vary accordingly. For instance:

If you have to make a frequent decision that affects the previous one you are dealing with a sequential DM problem. In those cases, reinforcement learning *2 or deep reinforcement learning *3 can be used to tackle this problem. Examples of these problems can be seen in video-games where the game AI needs to take different actions (policies) over time to maximise its score. (reward)

If you the problem is not sequential but you deal with multiple criteria to find the most attractive alternative then you are dealing with a multi-criteria decision-making problem, topic widely researched in operations research. There are some typically-used algorithms that are utilised to assist human-decision making like AHP*4, TOPSIS*5, ELECTRE*6, PROMETREE*7. An example of MCDC is selecting a house to buy, where you have to consider location, price among other desirable or undesirable characteristics.

Depending on the level of uncertainty, subjective data and incomplete information of the problem you might require to use fuzzy, intuitionistic or neutrosophic variations of the mentioned algorithms. *8

You might need to optimise DM through different competing goals. In that case, you are dealing with a multi-objective decision-making optimisation problem (MODM). See Decision trees*9, genetic algorithms*10 .

Furthermore, a DM problem can have different 'agents' making decision that can affect ours. So that is known as 'multi-agent' decision-making. In computer science, multi-agent system simulations are commonly used to research these problems. *11

You can also have the case where the agents have to make a collaborative decision that affects all of them. So that is known as 'group' decision-making.

In the industry, computational DM can be seen with the widely used recommender systems such as the ones in Netflix or Amazon.*13 In the B2B sector, AI in DM can be seen in decision-support systems and prescriptive analytics services *14.

I hope you find that information useful. There is indeed, much more about this complex topic, I just tried to summarise.

Some resources you might want to check:

• Deep RTS: A playground for reinforcement learning agents in real-time strategy game environments. (Repository: https://github.com/cair/deep-rts) (Pre-print Paper: https://arxiv.org/abs/1808.05032)
• OpenAI Gym: A general-purpose playground to test reinforcement learning AI algorithms. (Github: https://github.com/openai/gym, page: https://gym.openai.com/)
• DecisionRadar: An online application to apply TOPSIS decision-making algorithm. (Site: https://decision-radar.com/)
• AgentSimJS: A 3D multi-agent simulation system built in Javascript. (Repository: https://github.com/maxdeben83/agentsimjs)

REFERENCES:

• *1 Atkinson, J. W. (1964). An introduction to motivation.
• *1 Berridge, K. C. (2004). Motivation concepts in behavioral neuroscience. Physiology & behavior, 81(2), 179-209.
• *1 Hwang, C. L., & Yoon, K. (1981). Methods for multiple attribute decision making. In Multiple attribute decision making (pp. 58-191).Springer, Berlin, Heidelberg.
• *1 Tversky, A., & Kahneman, D.(1981). The framing of decisions and the psychology of choice science, 211(4481), 453-458.
• *2 Littman, M. L. (1994). Markov games as a framework for multi-agent reinforcement learning. In Machine Learning Proceedings 1994 (pp. 157-163).

• *3 Van Hasselt, H., Guez, A., & Silver, D. (2016, February). Deep Reinforcement Learning with Double Q-Learning. In AAAI (Vol. 2, p. 5).

• *4 Aczél, J., & Saaty, T. L. (1983). Procedures for synthesizing ratio judgements. Journal of Mathematical Psychology, 27(1), 93–102. doi:10.1016/0022-2496(83)90028-7

• *4 Saaty, R. W. (1987). The analytic hierarchy process—what it is and how it is used. Mathematical Modelling, 9(3-5), 167.
  doi:10.1016/0270-0255(87)90473-8

• *4 Saaty, T. L. (1986). Axiomatic Foundation of the Analytic Hierarchy Process. Management Science, 32(7), 841.
  doi:10.1287/mnsc.32.7.841

• *4 Hwang, C. L., & Yoon, K. (1981). Methods for multiple attribute decision making. In Multiple attribute decision making (pp. 58-191). Springer, Berlin, Heidelberg.

• *6 Zhou, Y. (1915). Multi-Criteria Decision Making in Software Development: A Systematic Literature Review.

• *7 Zhou, Y. (1915). Multi-Criteria Decision Making in Software Development: A Systematic Literature Review.

• *8 Pramanik, S., Biswas, P., & Giri, B. C. (2015). Hybrid vector similarity measures and their
  applications to multi-attribute decision making under neutrosophic environment. Neural Computing and Applications, 28(5), 1163
  doi:10.1007/s00521-015-2125-3

• *8 Mardani, A., Nilashi, M., Zavadskas, E. K., Awang, S. R., Zare, H., & Jamal, N. M. (2018). Decision Making Methods Based on Fuzzy Aggregation Operators: Three Decades Review from 1986 to 2017.
  International Journal of Information Technology & Decision Making, 17(02), 391–466. doi:10.1142/s021962201830001x

• *9 Zhao, H. (2007). A multi-objective genetic programming approach to developing Pareto optimal decision trees. Decision Support Systems, 43(3), 809-826.

• *9 Laumanns, M., & Ocenasek, J. (2002, September). Bayesian optimization algorithms for multi-objective optimization. In
  International Conference on Parallel Problem Solving from Nature (pp. 298-307). Springer, Berlin, Heidelberg.

• *9 Jin, Y. (Ed.). (2006). Multi-objective machine learning (Vol. 16). Springer Science & Business Media.

• 10 Tamaki, H., Kita, H., & Kobayashi, S. (1996, May). Multi-objective optimization by genetic algorithms: A review. In Evolutionary Computation, 1996., Proceedings of IEEE
  International Conference on (pp. 517-522). IEEE.

• *11 Rodriguez, S., Gaud, N., & Galland, S. (2014, August). SARL: a general-purpose agent-oriented programming language. In Web Intelligence (WI) and Intelligent Agent Technologies (IAT), 2014 IEEE/WIC/ACM International Joint Conferences on (Vol. 3, pp. 103-110). IEEE.

• *12 Rao, A. S. (1996, January). AgentSpeak (L): BDI agents speak out in a logical computable language. In European Workshop on Modelling Autonomous Agents in a Multi-Agent World (pp. 42-55). Springer, Berlin, Heidelberg.

• *13 Ricci, F., Rokach, L., & Shapira, B. (2015). Recommender systems: introduction and challenges. In Recommender systems handbook (pp. 1-34). Springer, Boston, MA.

• *14 https://www.ibm.com/analytics/prescriptive-analytics

Answer 3

http://msl.cs.uiuc.edu/~lavalle/cs397/

Read this. There are different ways to go about AI decision making. The examples are on the bottom.

Answer 4

There are simply too many examples to count! The whole (large) field of expert systems development, for example, consists of building programs whose explicit purpose is to emulate human decision making. Google "expert systems" to find thousands of examples.

I'll put in a plug for Jess, an expert system shell written in Java which I developed. It's used by thousands of companies world wide to automate decision-making processes.

Answer 5

I did a bit of browsing on the web and I came across this example project.

You might also check out the AI-Depot website here.

Probably the reason there are not so many working examples is because AI decision algorithms, such as neural networks, genetic algorithms, and decision trees get very complex, very quickly. Most developers of such algorithms, at least the algorithms that are stable and actually work, are very protective of their IP. And, for good reason.

In any case, hope this helped.

• Let It Be Known