Reference no: EM132750821
Long critical reflection report and demo of Jupiter Script via viva.
Tasks to be undertaken
The main task you are set is to expand the chatbot presented and fully explained in which he explores a fun and interesting use-case of recurrent sequence-to-sequence models.
You should use this code as the basis of your assignment and expand this system so that it becomes a practical (if still modest) piece of software. Modify it in any way you can think of that will make it more useful and robust. This is an open-ended project.
Possible modifications might include but are not limited to:
• Play and fine tune the RNN parameters.
• Develop ways to measure how accurate the responses from the bot are.
• Use other Neural Network algorithms including CNN and MLP
• Use other non AI techniques to implement the bot
• Explore the use of other data sets (look for them in the web, look at your whatsapp own private conversations, look at your facebook account etc.)
• Show some outputs (plots, stats) to be able to compare the different approaches
We will be looking for:
• Sensible extensions to the system.
• Sensible choice of dialogs.
• Sensible experiments towards fine tuning the system.
• Careful thought about the various design parameters for the system.
Deliverables to be submitted for assessment:
Deliverable 1:
Executive Summary
You will write a single A4 sheet (one page only) executive report
Deliverable 2:
Jupyter Script file
A separate Jupyter format lab file containing the Python source code should be submitted to the Blackboard separately. The file should have all the output results from running the scripts so that there is no need to re-run the script to see the outputs.
Deliverable 3:
Viva and demo of the system
A viva will be required after you submit your work. This will consist of a 5 minutes demo of your Jupyter project with 5 minutes for questions for a total of 10 minutes.
No time will be allowed to run or train the system.
As we only have 5 minutes for the demo, please make sure that all the code has been previously executed under Jupyter and that it displays the right outputs.
Enough evidence should be provided to show a series of conversations with the bot. Highlighting of the main contributions will also be required. Time slots for the presentation will be specified near the time.
Attachment:- Natural language processing.rar