Some links and stuff from around the web:
i. A lecture on Averaging algorithms and distributed optimization. He’s quite good but this is not for everyone; you need a maths/stats background to some extent to understand what’s going on. I’ve seen many types of lectures online, but this one is probably one of the ones ‘closest’ to the type of lectures that are available to students where I study the kind of stuff I study, in terms of the format; there’s a lot of math, there’s a very clearly defined structure and the lecturer knows exactly what he’s supposed to cover during the lecture, you proceed from the simple and then add some complexity/exceptions etc. along the way, some i’s and j’s will be mixed up and a plus or minus sign will need to be corrected somewhere, the lecturer rarely asks the people attending class any questions and if it’s a good lecture there will not be a lot of questions from the audience either. It reminded me of the econometrics lectures I had some time ago, also because the stuff covered in the lecture relates a bit to material covered back then (‘gradient-like methods’, the convergence properties of various optimization algorithms, etc.).
ii. Cyanide & happiness. I found the comic a week ago or so and I like it. A few examples (click to view full size):
iii. From edge.org: What is life? A 21st century perspective, by Craig Venter. Not a bad way to spend an hour of your life.
iv. A list of free statistical software available online. There are a lot of those around!
v. An awesome retraction-story. The peer-review process is not always bulletproof:
“[Hyung-In Moon] suggested preferred reviewers during the submission which were him or colleagues under bogus identities and accounts. In some cases the names of real people were provided (so if Googling them, you would see that they did exist) but he created email accounts for them which he or associates had access to and which were then used to provide peer review comments. In other cases he just made up names and email addresses. The review comments submitted by these reviewers were almost always favourable but still provided suggestions for paper improvement.” (via Ed Yong)
vi. “In a study now in press in Neurobiology of Aging (download PDF copy here), we studied the effects of healthy aging on how the brain processes different kinds of visual information. Based on prior work showing that visual attention towards objects predominantly recruited regions of the medial temporal lobe (MTL), compared to attention towards positions, we tested whether this specialization would wither with increasing age.
Basically, we tested the level of brain specialization by comparing the BOLD fMRI signal directly between object processing and position processing. We looked at each MTL structure individually by analyzing the results in each individual brain (native space) rather than relying on spatial normalization of brains, which is known to induce random and systematic distortions in MTL structures (see here and here for PDF of conference presentations I’ve had on this).
Running the test with functional MRI, we found that several regions showed a change in specialization. During encoding, the right amygdala and parahippocampal cortex, and tentatively other surrounding MTL regions, showed such decreases in specialization.
During preparation and rehearsal, no changes reached significance.
However, during the stage of recognition, more or less the entire MTL region demonstrated detrimental changes with age. That is, with increasing age, those regions that tend to show a strong response to object processing compared to spatial processing, now dwindle in this effect. At higher ages, such as 75+, the ability of the brain to differentiate between object and spatial content is gone in many crucial MTL structures.
This suggests that at least one important change with increasing age is its ability to differentiate between different kinds of content. If your brain is unable to selectively focus on one kind of information (and possibly inhibit processing of other aspects of the information), then neither learning or memory can operate successfully.” (link)
No comments yet.