Page 127 - Artificial Intellegence_v2.0_Class_12
P. 127

Reboot


                 Fill in the blanks.
                   .   he problem               methodology can be applied to real life problems as  ell.

                   .   istorical data in  hich the desired outcome is already kno n is called      .
                   .  In the modelling stage, the data scientist  ill construct      or               models.
                   .   he methodology for model building and deployment is an            process.






                       a  Brainy Fact

                  nvision, the a ard  inning iO  and Android smartphone app that allo s blind and visually impaired
                 people to independently access visual information around them, announced plans to integrate its AI
                 po ered soft are technology into  oogle  lass in  arch     .  he combination of  nvision's soft are
                 and  oogle  lass gives blind and visually impaired users a substantially less invasive and hands free
                 manner of accessing the  orld around them, giving them greater freedom and independence to
                 access and 'see' the  orld around them.



                                                                                                Experiential Learning

                        Video Session


                   can the    code or visit the follo ing link to  atch the video   o   nvision  orks
                  https      .youtube.com  atch v  eh  nq   o t    s
                  After  atching the video, ans er the follo ing question
                   o  does envision glasses help the visually impaired people










                   Model Validation


             here are mainly t o types of validation methods  hich are  rain  est  plit  valuation and  ross  alidation.  et us learn
            about them in detail.

            T r ain T est S plit Evaluation
             he train test procedure measures the performance of machine learning algorithms  hen they need to make predictions
            on data that  as not used to train the model.  hile it is quick and easy to use, it is only suitable for huge data sets.  he
            train test split cannot be utilised  hen the data sets are small and additional configurations are required, such as  hen
            the data set is not balanced.






                                                                                       C apstone  P roj e ct
   122   123   124   125   126   127   128   129   130   131   132