The Reviews Are In

Apparently the reviews were some time coming as one of the anonymous reviewers was ill and had only recently recovered. I can happily say that the reviews were worth the 9 month wait as, by and large, they were very positive. The only things I have left to do are: correct 20 small typos, clarify two definitions, use a more relevant citation for one topic, and discuss how my work relates to a method in statistics called “empirical Bayes” - something I had a suspicion was relevant but only really got my head around sometime after I submitted my original draft.

There were a number of other non-essential recommendations in the review which I’ll take into account but I’m pretty confident I can have the important ones knocked over this weekend.

Now, you’ll have to excuse my extreme lack of modesty while I cherry pick some of the very positive feedback I received. I’ve spent a aeon on this one piece of work so I’m going to milk these for all they’re worth:

On the Contributions:

• The thesis is exemplary in the field of ILP since it includes, in lock-step, theory, implementation and experimentation.
• … work that represents significant original contributions to the area of machine learning…
• … the approach provides a firm conceptual basis on which to develop further ideas.
• … the candidate has a clear and deep understanding of the mathematical concepts involved.
• … raises the state of the art in inductive transfer to a new level.
• … the approach and the results presented in this thesis are likely to become reference citations for any further research in this field.

On the Presentation:

• This dissertation is a pleasure to read.
• This thesis extremely well-written. The writing is lucid and yet clear and not over-formalized …
• The overall presentation of the thesis is of a high standard. … The writing is very readable and stands out in comparison with much of the field.
• It has been executed with skill, thoroughness, and completeness which makes this thesis one of the best pieces of graduate work I have seen in the recent years.

One reviewer even noticed the time I spent putting a thesis-long example together to help with the exposition noting that “it’s an excellent and non-trivial presentation tool and the candidate should be lauded for its use” and picked up on the “dash of discrete humour” I scattered through the beast to keep me sane.

Before I put my modesty back in, I’ll just have to finish with a very flattering concluding remark from one review:

To summarize, Mr. Reid’s thesis is an excellent, novel, high quality contribution to Machine Learning. This thesis definitely merits the award of the degree.

Ahhhh, catharsis… the gut-wrenching slog over the last many years is but a dim, dark memory.

Update: I’ve now finished the corrections so, behold: the compleat thesis! (2.3Mb PDF)