Jim Skinner (@JimRSkinner) header
Jim Skinner (@JimRSkinner) profile picture

Research Fellow working with @BreathSpec. Interested in machine learning, bayesian data analysis & data science.

Joined August 2015
Photo by Jim Skinner (@JimRSkinner)Photo by Jim Skinner (@JimRSkinner)Photo by Jim Skinner (@JimRSkinner)Photo by Jim Skinner (@JimRSkinner)Photo by Jim Skinner (@JimRSkinner)Photo by Jim Skinner (@JimRSkinner)Photo by Jim Skinner (@JimRSkinner)Photo by Jim Skinner (@JimRSkinner)Photo by Jim Skinner (@JimRSkinner)
Jim Skinner retweeted
Excellent comparison of regression and PCA in a visually pleasing gif. By @andre_quentin, ht @HernanBruno
4
408
26
1,280
Advice wanted: how should I prepare for my phd viva?
4
1
3
Start your week with a dose of CoreML! This Monday, we have @AymanBoustati leading a discussion on Normalising Flows. Join us at the usual time and place!
4
8
We are back again! Today we have another speaker series session with Virginia Aglietti, talking on about her research on making inference in multi-task Cox processes more efficient. Join us at the usual time and place!
2
5
Another week, another meeting! This week we have Michael Pearce answering questions about what it is like to work in @DeepMindAI . Join us for this exciting session!
1
8
ML Club speaker series back on! @JimRSkinner talking about smell and latent variable models! Join us on Monday 16:00 in D1.07 for biscuits... and knowledge
1
3
2
9
statisticians/"data scientists" of the world: What laptop to I want?
1
1
First blog post in a little while: "Smell-based disease diagnosis" jimskinner.github.io/post/sm… Some of the fun backstory behind the application of my PhD work. Nice and short!
1
4
I just submitted my Thesis on my birthday! Thank you for the amazing cake @IlianaSPeneva, and thank you everybody at @WarwickComplex for making the last 4 years amazing!
6
1
21
Jim Skinner retweeted
As technology becomes more human it may be able to better substitute for human relationships. In the meantime, services that offer human contact to the lonely will thrive. In Japan this manifests itself in agencies and apps that allow you to rent a family or a friend—
1
1
3
Jim Skinner retweeted
First 1e6 integers, represented as binary vectors indicating their prime factors, and laid out using the sparse matrix support in @leland_mcinnes's UMAP dimensionality reduction algorithm. This is from a 1000000x78628 (!) binary matrix. Very pretty structure emerges.
84
1,755
248
4,745
In LaTeX, if a page is >= 60% floats, it becomes a float-only page. IMHO, 60% is way too small. Change this to 80% with: \renewcommand{\floatpagefraction}{.8}
2
We identified a few unhealthy scenarios and preemptively named and addressed them: One is "butt in seat syndrome". It's when you sit in your chair until 5:30 or 6 or whenever other people leave even if you are totally burnt out. #DevDiscuss If you need to go home, go home!
2
17
3
82
It has taken me an entire PhD to appreciate how beautifully written David MacKay's book information theory book is.
1
15
Histograms are fantastic, but make sure your bin-width/number is chosen well. This is the _exact_ same data, plotted with different bin-widths. Notice that the pattern doesn't necessarily get clearer as bin num increases. #dataviz
18
253
27
619
Jim Skinner retweeted
The SVD provides insight into what L2-penalized linear regression does: it shrinks smaller principal components more (Hastie et al.) But what about a generalization to the exponential family (eg L2-penalized logistic regression)?
3
5
1
23
top
22
503
33
1,763
Jim Skinner retweeted
What great chance for @RedKnightLtd to promote the project in #Athens! all thanks to @EURACADEMY for organising this amazing event and @WyndhamAthens for hosting it at their great venue! look forward to the next one! #BreathSpec #H2020 #FTI #AMR #EURACADEMY #projectmanagment
2
3