• 'Galeotto' was my M.Sc. project and who supervised it! [Inferno, Canto V] From web-based search to K-Means... Baptism of the fire! Can you draw the N-Dimensional Space?
• Question: how about doing a PhD on more K-Means?
• Broken up with boyfriend who hated
  K-Means
• One year before starting my PhD, I heard these words: Why do you say "slow progress"? Your progress is NOT slow. The mere fact that we are considering an experiment in February means that we are going quite fast. Secondly, no one does a PhD on their own in this Department - that may be the case in, say, a CV department, but here, there is a lot of teamwork. Once you get over the initial technical issues, things will be much easier. I am SURE about this point. I still believe in these words.
• The PDF (probability density function) appeared. At the time, the only PDF I knew was the one from Adobe!
• Received email explaining 'convolution', the scariest thing I had ever seen. 2 thoughts came to my mind:
  (1) Forget the Phd (2) Go to the library. I followed the second one

  November 2004

• Started PhD at Imperial College in November 2004
• Happy to be back to London. Will work on my project. At last!
• Live in nice flat with porter close to Imperial. I am happy.

  Before Christmas 2004

• Created clever rules in outlook to automatically sort emails
• Filed wrong emails. Deleted rules!
• Supervisor teaching me basic maths. Feel ashamed.
• I have met the lie detection Professor. I was waiting for an hour. They forgot to tell him I was waiting! He saw me by chance... He waved. He then started going up and down from one floor to another of his University and from one room to another. From time to time he disappeared. No clue of what he was up to. I then managed to find him. He had finished to prepare his coffee! His room was full of papers everywhere. So much to talk about. At the end I left his building from the wrong door and set the fire alarm! PhD just started...
• Integration nightmare: I dreamt I was in an interval and couldn't get out
• Christmas dinner with PhD students. All strangely having dessert! Supervisor subsidied dinner
• Lie detectors are fascinating to play with, aren't they? ;-) - Was this comment deceptive? No idea

  January 2005

• Administration office got my registration wrong for the 5th time. Supervisor not surprised...
• Connected component labelling algorithm works! Supervisor happy

  February 2005

• Written the first algorithm that plots the intensity values of the hands! It works
• Biggest problem: to interpret results. Made a huge effort not to laugh in front of supervisor when results appeared on screen after running the 'pinv' command!
• Problem: I cannot help laughing when weird things happen in results (see PINV picture on the left). MUST NOT laugh and be professional
• Supervisor laughs more than me. I will be ok
• Long talks with Magic Circle to set up a lie detection experiment. Still invisible thoughts...
• I wrote loads of useless XML. I hope they will be useful later
• First day off since start of PhD: waited whole morning with unattractive people at Italian Embassy to sort out residence paper work...

  March 2005

• Maths driving me crazy. Left University in tears
• I only need enough maths every time to get by. I will survive. Talent or no talent there is no human language that I cannot learn and there is no mathematical concept that is beyond me!
• Printer full of decryption techniques. How embarassing!
• Spent a lot of time decrypting yet another message received. Anonymous author to be punished.
• Met lie detection Professor again. This time we talked about "manipulators" and "illustrators". How on earth will I be able to talk about these at Imperial?
• Serious human behaviour problem or at least I know it will be. Left University to think of a plan. Solution: I don't care!
• Supervisor said I am a "resilient creature"

  April and May 2005

• Worked really hard on a bayesian algorithm. It finds faces and hands! Unbelieavable
• I told a friend (well I think he is a friend) that it is hard to control a flirty situation. He replied: "Well, it depends on the people involved. (Are you generalising from a sample which is too small?" :-)) Deceptive...
• Started filming London characters. Had hard time with the one who was hanging his clothes by the river Thames to get them dry
• Completed a successful high-stakes experiment which, I hope, will make up for human behaviour problem
• First paper accepted! I have only started to understand content...

  June 2005

• There are few things you can do during invigilation:
  (1) If invigilating is in the right room, wait for loud drill to start and compute duration of sound until it stops. Calculate average duration
  (2) Carry on trying to do the questions on the exam paper with no success
  (3) Read paper that has been on your desk for more than a month
  (4) Count the number of strings on each desk at the start of the exam and make a prediction of what desks will have less strings at the end of the exam
  (5) Wait for the smell of food to hit you before you start the countdown to when students have to hand the exam papers (6-N) Ask Dimitris Parapadakis, the originator and owner of the idea, for the others
• Magic Circle has finally sent me letter to take exam. Don't have any props in London. Will study a show on ropes!

  July 2005

• Made a painting on covariance matrices.
• Supervisor keen to collaborate with ICA!
• Run brainstorming with magicians at Soho House. Supervisors had a good time. Magicians' ideas... vanished!
• One more person has joined my lie detection list: http://www.jiscmail.ac.uk/lists/LIE-DETECTION.html - Uuuuuuu!
• The world is not Gaussian
• Image pixels are not Gaussian
• With the right conditions they are. So the world is conditionally Gaussian
• Pluribus intentus, minor est ad singula sensus (Gaussian discussion over)
• Found a partner for doing photo/video trips at last!

  August 2005

• Add this to CV: excellent flirt management skills
• Spent days (and nights) rewriting tracking code. Still no results from experiments!
• Some people believe that to really test lie detection you have to use subjects who are not aware that they are forced to lie. I think this only applies to real-word applications. The more people are scared of being watched the more cues they are likely to generate...
• If you interview people who have committed a fairly high-stakes crime, then you may be breaking a lot of very serious institution regulations! The last thing I want is my supervisor or Imperial to be sued for breakdown of relationships. I can already see the headline: “Imperial experiment wrecked my marriage”
• Everybody lies.... every day, every hour, awake, asleep, in his dreams, in his joy, in his mourning. If he keeps his tongue still his hands, his feet, his eyes, his attitude will convey deception". Mark Twain - This quote is dedicated to one my friend who claims he never lies

  September 2005

• Right now I have a lot of weird data to process
• Spent 2 weeks on same velocity signals. Each time I extract the signals, they look different! Supervisor not impressed
• I always forget what the data is about and cause trouble to the subject! Subject upset

  October 2005

• First lab. It is possible to answer questions on things you don't know and see the students magically understand...
• People close to me got upset for no reason. Yet I am doing a PhD on human behaviour!
• I am broke. Will use a day-per-page pocket diary to write down what I spend. Writing down what I spend will make me spend less
• Partner for doing photo/video trips gone back to India :-(
• Second lab. Had to explain imaginary numbers!...
• Subject said there was a mistake in my truths/lies labels. I really do not want to know which one
• I keep dreaming that the subject's dog is chasing to attack me. Must start working on different data
• Third lab. Student curious to know how to plot infinity...
• Evolutionary theorists argue that our ancestors spent so much time cheating that we've developed "cheater detection" mechanisms in our brain that enable us to be intuitively suspicious
• Spent a life trying to detect the lies of the people around me. I have failed completely
• Successfully placed ad on "divorce aid" website to recruit liars
• While working hard on journal paper I forgot to work on transfer report!... Have to deliver report in 2 weeks!
• PANICK

  November 2005

• Must fill fridge with ready made meals to survive next 2 weeks while writing report
• One week over. Report rubbish. Supervisor not impressed
• Equations start making sense
• Wasted time doing useless drawings. Will keep them for final thesis...
• Report in good shape. I am SO surprised. Supervisor not surprised???
• Broke up with yet another non stimulating boyfriend
• Had to esplain scary convolution formula to M.Sc. students. Exciting
• Supervisor not keen to work with ICA
• Sold painting on covariance matrices to writer in Oxford

  December 2005

• My father in hospital. Must go to Italy. Forget the PhD
• Received message: I am getting curious whether there is enough common ground in our research interests and idea of friendship. Potential excellent subject for lie detection experiment!
• New "demystified math" book on my desk
• Potential excellent subject for lie detection experiment has changed mind!

      Go to Year 2006-2007 >>

From this:

"I am sure we'll be able to successfully market your product Mr. Ferguson. I have one of my best liars working on it right now..."

To this:

To more of this:

To this:

To this:

To this:

To more of this:

To this:

Subject: "After the experiment, I have dreamt I was killing my mother's dog"