Acquired Intelligence & Adaptive Behaviour Writing a lab report

Acquired Intelligence & Adaptive Behaviour

Writing a lab report

Christopher L. Buckley

Resources

Guide to Technical Report Writing

Ten simple rules for structuring papers

Brett Mensh and Konrad Kording

Slide material taken from

Reports not questions

In each lab you will get question sheet which you should work  through

However the portfolio piece will be delivered as a scientific report in the style of a scientific paper.

Portfolio

  • Lab report is about 8 pages in total, no penalisation for length but make it concise.
  • Append your code, only put pseudo algorithms in the main text.

Portfolio pieces

The courseworks are entitled

CW 1: Report on Neural Networks (not in portfolio)

CW 2: Report on Genetic Algorithms (portfolio contribution) CW 3: Report on Evolving controllers (portfolio contribution)

https://docs.google.com/spreadsheets/d/1fHCLeqpW2I3Xm-pXYHN97BUoLFbPGYzofFiMpclWuDg/edit?usp=sharing

Neural Networks

  1. Write a report comparing and contrasting information processing in Feedforward versus recurrent neural networks.

-Minimal required  technical achievements

Use figures and schematics to present you implementation of AND, OR and NOR gates (provide weight and bias values)

Present an attempt at hand designing or optimizing and XOR gate. Discuss the importance of the hidden layer.

Time/versus state plot for the recurrent neural network (RNN).

Present the dynamics for a fixed point, cyclic dynamics, and chaos.

  1. Present a quantitative investigation of how the number of layers effects the performance of a feedforward neural network

Minimal required  technical achievements

Train and present the results for an XOR gate, explain the importance of the hidden layer.

Examine how training performance (number of epochs)  changes with the number of layers.

  1. Select a topic of your choice, these could include looking at the impact on layers on the MNIST data set, or training in recurrent neural networks or another topic, please contact me on Discord to check if it is okay.

Portfolio

Technical (30): The quality of your code and the algorithms you present.

Presentation (20): The quality of writing and organisation of the submitted document, the quality of the figures and diagrams.

Context (20): The extent to which you have motivated the work and discussed the results in the context of the ideas presented in the lectures.

Research (30): The extent to which you’ve gone beyond the lecture material and brought in ideas from the course reading, from other sources and your own ideas.

Typical Report Structure

  • Title page
  • Abstract/summary
  • Introduction
  • Methods
  • Results
  • Analysis and discussion
  • Summary and conclusions
  • References/bibliography
  • Appendices

Lab Report: Component Parts

Abstract

Introduction

Methods

Results

Discussion

Conclusion

This is the order in which you read lab report

Not the order in which you write it!

Writing Process

  • Start with the data – not the introduction
  • Narrow them down to a few figures
  • Assemble them into a story board
  • Find the trends in the figures. Find the one thing that ties them together
  • Tell your readers how to read your figures and what the main point is
  • Then map out the story that tells what the main point is

Writing Process

  • Start with Methods and Results sections
  • Connect results with how you got them
  • Then connect your interpretation of results (Discussion) to scientific assumptions or principles (Theory)
  • Connect what you set out to do (Introduction) to what you found (Conclusion)

Writing tips

  • The report should read like a starting from motvation ending with what you found.

Your write retrospectively and  make it seem that you always knew where you were going.

Writing Process – stages

  • Planning

○ Purpose of section

○ Brainstorm, mindmap, outline

  • Writing
  • Revising
  • Submit!

Technical Communication- 5Cs

  1. Clarity
  2. Conciseness
  3. Concreteness
  4. Coherence
  5. Context

Writing a Lab Report

Introduction

  • Provides background information (e.g., previous studies) and includes the objectives and the hypothesis.

Method

  • Specifies the details of your study, procedures or related analytic tools. In other words, how did you do it?

Results

  • Reports the data and analyses based on the data. State whether the results were consistent with the hypotheses, usually without interpretation of any wider meaning or importance. What did you find?

Discussion

  • Interprets the results of the experiment in terms of wider meaning and importance. What do the results mean?

References ● List of works used to write the lab report.

http://www.sussex.ac.uk/ei/internal/forstudents/engineeringdesign/studyguides/techreportwriting#1

Abstract

Introduction Section

“The introduction states the objective or purpose of the work and provides the reader with important background and/or theory to the experiment.”

Writing prompts for the introduction

  • What kind of problem did you work on?
  • Why did you work on this problem?
  • What should the reader know or understand when he/she is finished reading the report?

Writing prompts for introduction

  • Which research question did you set out to answer?
  • What was your expected answer or assumptions about the outcome of this investigation?
    • Hypothesis?

○ Designed to prove?

  • Relate assumptions to findings

Introduction Section

The introduction should be self-contained.

  • The question.
  • What you found.

Methods Section

  • Accurate and complete account of what you did and what methods you used
  • Usually a chronological structure but
  • Write in past tense

Methods Section

  • This gives an overview of how you investigated the questions you motivate in the introduction i.e schematics of the networks and pseudo algorithms.
  • How much detail? Imagine someone tried to recreate your work with only your report document and the question sheet alone. Could they do it? If they couldn’t then you need to put in some more details.

Results Section

  • Present data
  • State in verbal form as well as visual
  • Use sentence to draw attention to key points in graphs, figures, etc.
  • Number and title tables and graphs
  • Use appendix for raw data or complex calculations

Writing prompts for Results

  • What are your results?
  • Is the data presented so results are clear, logical and self-explanatory?
  • What is the main point – what ties results together?
  • The order

Writing prompts for Results

  • The order is driven by the resutls i.e a prior set results motivates the next set.

Discussion Section

  • “You show that you understand the experiment beyond the simple level of completing it.”
    • Explain

○ Analyse

○ Interpret

Discussion Section

  • A self-critical look at what you have achieved, and what you failed to achieve.
  • How does it fit with the ideas that you motivated in the introduction
  • Interprets the results of the experiment in terms of wider meaning and importance

Writing prompts for discussion section

  • Analysis
    • What do the results indicate clearly?

○ What are the sources of error?

○ How do the results compare to the theory/hypothesis? ● Interpretation

○ What is the significance of the results?

○ How do you justify that interpretation?

○ Suggested improvements for future research?

Conclusion

  • Usually short in student lab reports
  • State what you know as result of lab
  • No new information

Conclusion

  • Summarise what you found and describe future work and studies

End matter

  • Then a bibliography, properly referenced.
  • And an Appendix with all your code, properly presented and commented.
  • N.B. remember to spell-check and proof-read!

Structure

  • Can’t change component parts
  • But can
    • Make interesting and readable by focus on internal structure of sections

○ Way sections flow together

○ What info included, left out, emphasized ● Report tells a story!

Required content for portfolio pieces

The required context for each portfolio piece is listed on the Canvas website.

However to get the highest mark you will be expect to have gone beyond this in direction you choose.

Things to remember

  • No screenshots. Export figures from matlab.
  • Axes labels.
  • Think about putting more than one data line on each graph: better for comparisons.
  • Make pseudocode short: not full code
  • Introduction needs to link to what you do in your report