The Body Fat Setpoint, Part IV: Changing the Setpoint

Prevention is Easier than Cure

Experiments in animals have confirmed what common sense suggests: it's easier to prevent health problems than to reverse them. Still, many health conditions can be improved, and in some cases reversed, through lifestyle interventions. It's important to have realistic expectations and to be kind to oneself. Cultivating a drill sergeant mentality will not improve quality of life, and isn't likely to be sustainable.

Fat Loss: a New Approach

If there's one thing that's consistent in the medical literature, it's that telling people to eat fewer calories isn't a very effective fat loss strategy, despite the fact that it works if strictly adhered to. Many people who use this strategy see transient fat loss, followed by fat regain and a feeling of defeat. There's a simple reason for it: the body doesn't want to lose weight. It can be difficult to fight the fat mass setpoint, and the body will use every tool it has to maintain its preferred level of fat: hunger, increased interest in food, reduced body temperature, higher muscle efficiency (i.e., less energy is expended for the same movement), lethargy, lowered immune function, et cetera.

Therefore, what we need for sustainable fat loss is not starvation; we need a treatment that lowers the fat mass setpoint. There are several criteria that this treatment will have to meet to qualify:

1. It must cause fat loss
2. It must not involve deliberate calorie restriction
3. It must maintain fat loss over a long period of time
4. It must not be harmful to overall health

I also prefer strategies that make sense from the perspective of human evolution.

Strategies: Diet Pattern

One treatment that fits my criteria is low-carbohydrate dieting. Overweight people eating low-carbohydrate diets generally lose some fat and spontaneously reduce their calorie intake. In fact, in several diet studies, investigators compared an all-you-can-eat low-carbohydrate diet with a calorie-restricted low-fat diet. The low-carbohydrate dieters generally reduced their calorie intake and body fat to a similar or greater degree than the low-fat dieters, despite the fact that they ate all the calories they wanted (1). This may suggest that their fat mass setpoint had changed. At this point, I think moderate carbohydrate restriction may be preferable to strict carbohydrate restriction for some people, due to the increasing number of reports I've read of people doing poorly in the long run on extremely low-carbohydrate diets.  Furthermore, controlled trials of low-carb diets show that the long-term weight loss, despite being greater than low-fat diets, is not that impressive for the "average person".  Some people find it highly effective, while most people find it moderately effective or even ineffective.

Another strategy that appears preferable is the "paleolithic" diet. In Dr. Staffan Lindeberg's 2007 diet study, overweight volunteers with heart disease lost fat and reduced their calorie intake to a remarkable degree while eating a diet consistent with our hunter-gatherer heritage (3). This result is consistent with another diet trial of the paleolithic diet in diabetics (4). In post hoc analysis, Dr. Lindeberg's group showed that the reduction in weight was apparently independent of changes in carbohydrate intake*. This suggests that the paleolithic diet has health benefits that are independent of carbohydrate intake.

Strategies: Gastrointestinal Health

Since the gastrointestinal (GI) tract is so intimately involved in body fat metabolism and overall health (see the former post), the next strategy is to improve GI health. There are a number of ways to do this, but they all center around four things:

1. Don't eat food that encourages the growth of harmful bacteria
2. Eat food that encourages the growth of good bacteria
3. Don't eat food that impairs gut barrier function
4. Eat food that promotes gut barrier health

The first one is pretty easy in theory: avoid fermentable substances of which you're intolerant.  This can include lactose (milk) and certain polysaccharides, and a number of other FODMAPs.  For the second and fourth points, make sure to eat fermentable fiber. In one trial, oligofructose supplements led to sustained fat loss, without any other changes in diet (5). This is consistent with experiments in rodents showing improvements in gut bacteria profile, gut barrier health, glucose tolerance and body fat mass with oligofructose supplementation (6, 7, 8).  However, oligofructose is a FODMAP and therefore will be poorly tolerated by a subset of people.

The colon is packed with symbiotic bacteria, and is the site of most intestinal fermentation. The small intestine contains fewer bacteria, but gut barrier function there is critical as well. The small intestine is where the GI doctor will take a biopsy to look for celiac disease. Celiac disease is a degeneration of the small intestinal lining due to an autoimmune reaction caused by gluten (in wheat, barley and rye). This brings us to one of the most important elements of maintaining gut barrier health: avoiding food sensitivities. Gluten and casein (in dairy protein) are the two most common offenders. Gluten sensitivity is more common than most people realize; just under 1% of young adults and the prevalence increases with age.

Eating raw fermented foods such as sauerkraut, kimchi, yogurt and half-sour pickles also helps maintain the integrity of the upper GI tract. I doubt these have any effect on the colon, given the huge number of bacteria already present.

Strategies: Miscellaneous

Anecdotally, many people have had success using intermittent fasting (IF) for fat loss. There's some evidence in the scientific literature that IF and related approaches may be helpful (14). There are different approaches to IF, but a common and effective method is to do two complete 24-hour fasts per week. It's important to note that IF isn't about restricting calories, it's about resetting the fat mass setpoint. After a fast, allow yourself to eat quality food until you're no longer hungry.

Insufficient sleep has been strongly and repeatedly linked to obesity. Whether it's a cause or consequence of obesity I can't say for sure, but in any case it's important for health to sleep until you feel rested. If your sleep quality is poor due to psychological stress, meditating before bedtime may help. I find that meditation has a remarkable effect on my sleep quality. Due to the poor development of oral and nasal structures in industrial nations, many people do not breathe effectively and may suffer from conditions such as sleep apnea that reduce sleep quality. Overweight also contributes to these problems.


* Since reducing carbohydrate intake wasn't part of the intervention, this result is observational.

it's the weekend....

be girly.

*images courtesy of v., coco and kelley, eileen kathryn boyd, raina kattelson, mrs. blandings

On MRF factorization

Recently, I figured out I didn't understand two simple things about MRFs:

1. When we are talking about MRF factorization, in the formulation of the likelihood

   
   should we consider all the cliques in the network or only maximal ones?
2. Generally, why do we have a right to factorize the likelihood this way?

I asked some folks, and it turned out that nobody understood this basic property clearly (In spite of Dmitry Vetrov assured me he had explained that fact in his courses). Finally, I've found the answers in Christopher Bishop's book. If you are not familiar with Markov Random Fields or graphical models, I recommend you to read the book, it is relatively understandable.

So, let's start with the second question. For those of you who like names, it is exactly the necessity part of the Hammersley-Clifford theorem. As Bishop put it:

If we consider two nodes x_i and x_j that are not connected by a link, then these variables must be conditionally independent given all other nodes in the graph. This follows from the fact that there is no direct path between the two nodes, and all other paths pass through nodes that are observed, and hence those paths are blocked. This conditional independence property can be expressed as

where x_\{i,j} denotes the set x of all variables with x_i and x_j removed. The factorization of the joint distribution must therefore be such that x_i and x_j do not appear in the same factor in order for the conditional independence property to hold for all possible distributions belonging to the graph.

One could find the answer to the first question in Bishop's book as well. Normally, we should include all the cliques to the product. Actually, every function of variables X is also a function of any Y that is a superset of X. Based on that, a potential for a non-maximal clique could be merged into a potential for any maximal clique such that it is a supergraph of the first clique! Thus, both formulations (with all cliques and only maximal cliques) are equally expressive.

camouflage

there is a paint color by ben moore that i am totally in love with called camouflage (2143-40). it is a beautiful soft green without screaming "green". i used it on the exterior of a beach house along with the accent color, alexandria beige (HC-77). these images remind me of how pretty this color is.

*images courtesy of mod girl, martha stewart living, house and home, southern accents

"me time"

have you ever suffered from pregnancy meltdowns (where you turn completely crazy for no apparent reason)? i had one last night....it was caused by the library calling to tell me to return "elmo's potty time" for the second time. i sat on the couch and shed some tears and said i need some "me time". it lasted about ten minutes...then i felt silly and ate ice cream.

ha....being pregnant- isn't it grand :)

*images courtesy of coco cozy, house of turquoise/atlanta home magazine, real simple (bottom two images)

The Body Fat Setpoint, Part III: Dietary Causes of Obesity

[2013 update: I've edited this post to remove elements that I feel were poorly supported.  I now think that changes in the setpoint are at least partially secondary to passive overconsumption of calories, particularly low quality calories]

What Caused the Setpoint to Change?

We have two criteria to narrow our search for the cause of modern fat gain:

1. It has to be new to the human environment
2. At some point, it has to cause leptin resistance or otherwise disturb the setpoint

Although I believe that exercise is part of a healthy lifestyle, and can help prevent fat gain and to some degree treat overweight, it probably can't explain the recent increase in fat mass in modern nations. This is because exercise doesn't appear to have declined. There are various other possible explanations, such as industrial pollutants, a lack of sleep and psychological stress, which may play a role. But I feel that diet is likely to be the primary cause. When you're drinking 20 oz Cokes, bisphenol-A contamination is the least of your worries.

In the last post, I described two mechanisms that may contribute to elevating the body fat set point by causing leptin resistance: inflammation in the hypothalamus, and impaired leptin transport into the brain due to elevated triglycerides. After more reading and discussing it with my mentor, I've decided that the triglyceride hypothesis is on shaky ground*. Nevertheless, it is consistent with certain observations:

• Fibrate drugs that lower triglycerides can lower fat mass in rodents and humans
• Low-carbohydrate diets are somewhat effective for fat loss and lower triglycerides
• Fructose can cause leptin resistance in rodents and it elevates triglycerides (1)
• Fish oil reduces triglycerides. Some but not all studies have shown that fish oil aids fat loss (2)

Inflammation in the hypothalamus, with accompanying resistance to leptin signaling, has been reported in a number of animal studies of diet-induced obesity. I feel it's likely to occur in humans as well, although the dietary causes are probably different for humans. The hypothalamus is the primary site where leptin acts to regulate fat mass (3). Importantly, preventing inflammation in the brain prevents leptin resistance and obesity in diet-induced obese mice (3.1). The hypothalamus is likely to be the most important site of action. Research is underway on this.

The Role of Digestive Health

What causes inflammation in the hypothalamus? One of the most interesting hypotheses is that increased intestinal permeability allows inflammatory substances to cross into the circulation from the gut, irritating a number of tissues including the hypothalamus.

Dr. Remy Burcelin and his group have spearheaded this research. They've shown that high-fat diets cause obesity in mice, and that they also increase the level of an inflammatory substance called lipopolysaccharide (LPS) in the blood. LPS is produced by gram-negative bacteria in the gut and is one of the main factors that activates the immune system during an infection. Antibiotics that kill gram-negative bacteria in the gut prevent the negative consequences of high-fat feeding in mice.

Burcelin's group showed that infusing LPS into mice on a low-fat chow diet causes them to become obese and insulin resistant just like high-fat fed mice (4). Furthermore, adding 10% of the soluble fiber oligofructose to the high-fat diet prevented the increase in intestinal permeability and also largely prevented the body fat gain and insulin resistance from high-fat feeding (5). Oligofructose is food for friendly gut bacteria and ends up being converted to butyrate and other short-chain fatty acids in the colon. This results in lower intestinal permeability to toxins such as LPS. This is particularly interesting because oligofructose supplements cause fat loss in humans (6).

A recent study showed that blood LPS levels are correlated with body fat, elevated cholesterol and triglycerides, and insulin resistance in humans (7). However, a separate study didn't come to the same conclusion (8). The discrepancy may be due to the fact that LPS isn't the only inflammatory substance to cross the gut lining-- other substances may also be involved. Anything in the blood that shouldn't be there is potentially inflammatory.

Overall, I think gut dysfunction could play a role in obesity and other modern metabolic problems.
Exiting the Niche

I believe that we have strayed too far from our species' ecological niche, and our health is suffering. One manifestation of that is body fat gain. Many factors probably contribute, but I believe that diet is the most important. A diet heavy in nutrient-poor refined carbohydrates and industrial omega-6 oils, high in gut irritating substances such as gluten and sugar, and a lack of direct sunlight, have caused us to lose the robust digestion and good micronutrient status that characterized our distant ancestors. I believe that one consequence has been the dysregulation of the system that maintains the fat mass "setpoint". This has resulted in an increase in body fat in 20th century affluent nations, and other cultures eating our industrial food products.

In the next post, I'll discuss my thoughts on how to reset the body fat setpoint.

* The ratio of leptin in the serum to leptin in the brain is diminished in obesity, but given that serum leptin is very high in the obese, the absolute level of leptin in the brain is typically not lower than a lean person. Leptin is transported into the brain by a transport mechanism that saturates when serum leptin is not that much higher than the normal level for a lean person. Therefore, the fact that the ratio of serum to brain leptin is higher in the obese does not necessarily reflect a defect in transport, but rather the fact that the mechanism that transports leptin is already at full capacity.

On image labelling

Labelling data is a labourous side task that arises in most computer vision projects. Since the developers usually don't want to spend their time for such a dumb work, there exist a number of workarounds. Let me enumerate some I've heard of:

1. At Academia, the task of labelling is usually being endured on [PhD] students' broad shoulders. The funny part is the students are not always enrolled in the relevant project. At Graphics & Media Lab, students who have not attended enough seminars by the time of revision, should label some data sets for the lab projects.
2. One could also hire some people to label her data. Since the developers/researchers are relatively high-paid, it is economic to hire other folks (sometimes, they are students as well). UPDATE: hr0nix mentioned in the comment that there exists the Mechanical Turk service that helps requesters to find contractors.
3. The more witty way is to use applied psychology. For example, Google transformed the labelling process to the game. During the gameplay, you and your randomly chosen partner tag images. Sooner you tag an image with the same tag, more points you get. The brilliant idea! Believe or not, when I first saw it, I was carried away and could not stop playing until my friends dragged me out for a pizza!
4. The most revolutionary approach was introduced by Densey Tan. Here is a popular explanation of what he has done. The idea is to capture labels straight from one's brain using EEG/fMRI/whatnot. Now they can perform only 2 or 3 class labelling, but (I hope) it is only the beginning.

The last point reminds me my old thoughts about the future of machine learning (or at least ML applied to Vision). Nowadays we deal with ensembles of weak classifiers, such as decision trees, stamps etc. One can use guinea pigs as weak classifiers! I suppose their brain is developed enough to understand 3d structure of the scene in the way human brain does, while modern computer vision systems lack for this ability. The animals are to be learned, for example, by experiencing an electric shock in case of wrong answers. Now, it is not obligatory to train "experts", it is sufficient to analyse their brain activity. Isn't it a breakthrough? :)

OpenCV bindings

If you are a computer vision researcher or an engineer, you cannot miss OpenCV library. Even if you are not, it could be useful to you. For example, here is a funny application: camera shots a programmer's face when a merge fails. If you are not familiar with the library, I recommend you to look through the list of its features on Wikipedia.

OpenCV is written in C to be extremely portable (for example, to DSP). The fact is C is not very popular nowadays. The recent release 2.0 contains (besides the other decent stuff) also C++ and Python wrappers. What about the other languages?

There are a number of C# wrappers. The most known is EmguCV, which is reported to be the only C# wrapper that supports OpenCV 2.0 (actually, I don't now what it means, but I suppose the API should correspond the C++ interface). It is distributed under GPL or the "Commercial License with a small fee".

As for Java, JavaCV seems to be the only viable wrapper. It also contains wrappers for other popular libraries like FFmpeg. It is also distributed under GPL, but the author promised to discuss weakening it if needed.

OpenCV was being supported by Intel, but it became a FOSS project recently. They are also going to participate in Google Summer of Code. If they will succeed, you might try to apply. I think it is a nice experience to develop such a popular library and be paid for it. :)

One of the fields they want to develop is augmented reality support for Android operating system. When I get known that there is an AR API in Android, I decided to try it. So, this is going to be a good opportunity!

UPD (Aug 6, 2011). There appeared a Haskell (!) wrapper for OpenCV by Noam Lewis.

Also, OpenCV folks are developing the official Java wrapper. Looking forward to use it!

just because

i really like these images....

i'm off to get my weekly shot. it's loads of fun but it's keeping my little man safe and sound and hopefully (fingers crossed) preventing me from going on bed rest.

*images courtesy of aubrey road/ martha stewart, everything fabulous

Krauss's New Article on Saturated Fat Intervention Trials

Dr. Ronald Krauss's group just published another article in the American Journal of Clinical Nutrition, this time on the intervention trials examining the effectiveness of reducing saturated fat and/or replacing it with other nutrients, particularly carbohydrate or polyunsaturated seed oils. I don't agree with everything in this article. For example, they cite the Finnish Mental Hospital trial. They openly acknowledge some contradictory data, although they left out the Sydney diet-heart study and the Rose et al. corn oil study, both of which suggested increased mortality from replacing animal fats with polyunsaturated seed oils. Nevertheless, here is the conclusion:

Particularly given the differential effects of dietary saturated fats and carbohydrates on concentrations of larger and smaller LDL particles, respectively, dietary efforts to improve the increasing burden of CVD risk associated with atherogenic dyslipidemia should primarily emphasize the limitation of refined carbohydrate intakes and a reduction in excess adiposity.

*sigh*

i'm not typically one to hang out in a field w/ just a blazer and some rain boots on, but this picture is just about the cutest thing ever....and makes me long for spring....

*images courtesy of sweetie pie pumpkin noodle, tracery interiors, design sponge, porters

for those interested....

i have officially started the designer's attic again......maybe i'll see some of y'all over there :)

today's deal: ralph lauren wallpaper....

*images courtesy of the desk of annie, the designer's attic

Canadians Break Records at H.U.R.T. 100

100 Mile Finishers:
-Gary Robbins North Vancouver BC 20:12 1st Overall; new CR (photo)
-Tracy Garneau Veron BC 24:06 3rd Overall; 1st Female; new CR (photo)
-Jamie Nott Calgary AB 31:40 17th Overall
-Monica Scholz Jerseyville ON 32:02 18th Overall; 2nd Female
-Matthew Session Burnaby BC 32:10 19th Overall
-Charlotte Vasarhelyi Breslau ON 32:28 22nd Overall; 3rd Female
-Tasha Nott Calgary, ON 35:50 46th Overall; 5th Female

100K Finishers:
-Alan Haner Mindemoya ON
-John Turner Burlington ON
-Tammy Sieminowski Toronto ON
-Phillip McColl Jerseyville ON
-Sharon Zelinki Toronto ON
-Jennifer-Anne Meneray London ON

Computer Vision: Fact & Fiction

I was surfing the web today and came upon Stanford CS 223B course (Introduction to Computer Vision), which is said to be fucking hard. The first course homework is to watch the series of films "Computer Vision: Fact & Fiction" where computer vision stars (like David Forsyth and Andrew Zisserman) analyse computer vision technologies featured in Hollywood movies. The videos require no background in Vision and might be interesting to everyone. To me, it is also interesting to see how the famous vision folks look and talk.

My friend Tolya Yudanov spoke about that to talk about realistic in The Terminator movie is like "arguing about physical correctness of animé. Terminator is the complex AI of the future, and it is stupid to apply modern computer vision criteria to it." So, it is a good illustration of the concept of computer blindness. I encourage you to watch the videos, they are worth watching.

The Body Fat Setpoint, Part II: Mechanisms of Fat Gain

The Timeline of Fat Gain

Modern humans are unusual mammals in that fat mass varies greatly between individuals. Some animals carry a large amount of fat for a specific purpose, such as hibernation or migration. But all individuals of the same sex and social position will carry approximately the same amount of fat at any given time of year. Likewise, in hunter-gatherer societies worldwide, there isn't much variation in body weight-- nearly everyone is lean. Not necessarily lean like Usain Bolt, but not overweight.

Although overweight and obesity occurred forty years ago in the U.S. and U.K., they were much less common than today, particularly in children. Here are data from the U.S. Centers for Disease Control NHANES surveys (from this post):

Together, this shows that a) leanness is the most natural condition for the human body, and b) something about our changing environment, not our genes, has caused our body fat to grow.

Fat Mass is Regulated by a Feedback Circuit Between Fat Tissue and the Brain

In the last post, I described how the body regulates fat mass, attempting to keep it within a narrow window or "setpoint". Body fat produces a hormone called leptin, which signals to the brain and other organs to decrease appetite, increase the metabolic rate and increase physical activity. More fat means more leptin, which then causes the extra fat to be burned. The little glitch is that some people become resistant to leptin, so that their brain doesn't hear the fat tissue screaming that it's already full. Leptin resistance nearly always accompanies obesity, because it's a precondition of significant fat gain. If a person weren't leptin resistant, he wouldn't have the ability to gain more than a few pounds of fat without heroic overeating (which is very very unpleasant when your brain is telling you to stop). Animal models of leptin resistance develop something that resembles human metabolic syndrome (abdominal obesity, blood lipid abnormalities, insulin resistance, high blood pressure).

The Role of the Hypothalamus

The hypothalamus is on the underside of the brain connected to the pituitary gland. It's the main site of leptin action in the brain, and it controls the majority of leptin's effects on appetite, energy expenditure and insulin sensitivity. Most of the known gene variations that are associated with overweight in humans influence the function of the hypothalamus in some way (1). Not surprisingly, leptin resistance in the hypothalamus has been proposed as a cause of obesity. It's been shown in rats and mice that hypothalamic leptin resistance occurs in diet-induced obesity, and it's almost certainly the case in humans as well. What's causing leptin resistance in the hypothalamus?

There are three leading explanations at this point that are not mutually exclusive. One is cellular stress in the endoplasmic reticulum, a structure inside the cell that's used for protein synthesis and folding. I've read the most recent paper on this in detail, and I found it unconvincing (2). I'm open to the idea, but it needs more rigorous support.

A second explanation is inflammation in the hypothalamus. Inflammation inhibits leptin and insulin signaling in a variety of cell types. At least two studies have shown that diet-induced obesity in rodents leads to inflammation in the hypothalamus (3, 4)*. [2013 update: several studies have shown that preventing hypothalamic inflammation attenuates fat gain in obesity models].  If leptin is getting to the hypothalamus, but the hypothalamus is insensitive to it, it will require more leptin to get the same signal, and fat mass will creep up until it reaches a higher setpoint.

The other possibility is that leptin simply isn't reaching the hypothalamus. The brain is a unique organ. It's enclosed by the blood-brain barrier (BBB), which greatly restricts what can enter and leave it. Both insulin and leptin are actively transported across the BBB. It's been known for a decade that obesity in rodents is associated with a lower rate of leptin transport across the BBB (5, 6).

What causes a decrease in leptin transport across the BBB? Triglycerides are a major factor. These are circulating fats going from the liver and the digestive tract to other tissues. They're one of the blood lipid measurements the doctor makes when he draws your blood. Several studies in rodents have shown that high triglycerides cause a reduction in leptin transport across the BBB, and reducing triglycerides allows greater leptin transport and fat loss (7, 8). In support of this theory, the triglyceride-reducing drug gemfibrozil also causes weight loss in humans (9)**. Low-carbohydrate diets, and avoiding sugar and refined carbohydrates in particular, reduce triglycerides and produce weight loss, although that doesn't necessarily mean one causes the other.

In the next post, I'll get more specific about what factors could be causing hypothalamic inflammation and/or reduced leptin transport across the BBB. I'll also discuss some ideas on how to reduce leptin resistance sustainably through diet and exercise.


* This is accomplished by feeding them sad little pellets that look like raw cookie dough. They're made up mostly of lard, soybean oil, casein, maltodextrin or cornstarch, sugar, vitamins and minerals (this is a link to the the most commonly used diet for inducing obesity in rodents). Food doesn't get any more refined than this stuff, and adding just about anything to it, from fiber to fruit extracts, makes it less damaging.

** Fibrates are PPAR agonists, so the weight loss could also be due to something besides the reduction in triglycerides.

it's the weekend....

don't work late, enjoy a fire, have a dinner party w/ friends, sleep in on saturday, eat breakfast in bed, pamper yourself to a long shower....

whatever you do, enjoy it! and i'll see you next week :)

*images courtesy of coco cozy, high street market blog, style swoon, bippity boppity boo, katy elliot

The disruption of scholarly publishing [in Russia]

You know, I am relatively new to the big science. The first 2-column paper I read was Antonin Guttman's paper about R-Trees.¹ It was about 2.5 years ago. So, I've never used published journals or conference proceedings. I have been wondering why do they spend money to printing journals if the researchers usually publish their papers on their home pages. Recently, I've read the article about the disruption of scholarly publishing by Michael Clarke. The article is quite dragged out, so I summarize its ideas in the following paragraphs.

When Tim Berners-Lee came up with the idea of WWW in 1991, he thought of its purpose as some kind of scientific media, which would replace conferences and journals. Nowadays, the Internet is used for social networking, illegal distribution of music and pornography, but we still have journals and conferences off-line. Why?

The author indicates 5 main points:

• Dissemination - journals distribute papers all over the world
• Registration of discovery - to know who was the first, Popov or Marconi
• Validation - to ensure that the results are correct; now provided by peer-review
  
• Filtration - you are likely to read a paper from the journal with bigger impact factor
  
• Designation - if you have publications on top conferences, you might be a cool scientist

The author argues that only the last point is critical for the modern system of journals and conferences. The others could be better handled with on-line services, we already have a lot of examples.

But in Russia, we don't rank scientists according to their citation index!² (because we don't have one :) ) So, the designation problem is not solved here, and Russia is already ready to the new publishing system. Why do we still have journals? I don't know, probably the sluggishness of minds...

¹ Actually, I had read this paper about Google before, but it accidentally was not 2-column formatted. :)
² I am cunning a bit: some journals carry out designation purposes. I mean journals published by VAK. One should have at least one publication in such a journal to get Candidate of Science degree (Russian PhD equivalent). But the review process is usually lousy, there was also Rooter case with one of them.

New Saturated Fat Review Article by Dr. Ronald Krauss

Dr. Ronald Krauss's group has published a review article titled "Meta-analysis of prospective cohort studies evaluating the association of saturated fat with cardiovascular disease". As anyone who's familiar with the literature could have predicted (including myself), they found no association whatsoever between saturated fat intake and heart disease or stroke:

A meta-analysis of prospective epidemiologic studies showed that there is no significant evidence for concluding that dietary saturated fat is associated with an increased risk of CHD or CVD.

love....

the contrast you get with a dark grout and a classic, white subway tile.

*images courtesy of torie jayne, city sage, marley and lockyer, inspiration for everyday living

Testing machine learning algorithms

Tests are the only way to estimate the quality of a machine learning algorithm in practice. In order to prove your algorithm is usable, you need to design good tests. To design test you should collect the data and split them into the train set and the test set.

Machine learning theorists say that the train and the test sets should come from the single probability distribution (unless they are talking about transfer learning or some kinds of on-line learning). But in practice it is a bit more complicated. We are currently working on the problem of laser scan points classification. It is not a trivial task to design tests! We have scans from different domains (aerial scans, scans from moving vehicles, stationary terrestrial scans), and for each domain we would like to have kind of universal benchmark. It means that a wide range of algorithms are supposed to be tested with the test, so the test may not stimulate overfitting.

So, how can we split the data? To satisfy the claim of the single distribution, we can add the odd points from the cloud to the train set and even points to the test set. This is a bad idea. Suppose your classifier use 3D coordinates of a point as the features. For each point in the test set, we have a similar point in the train set. Therefore we get nearly 100% precision using such a primitive learner. Such benchmark is not enough challenging.

Well, let's split every scan into a few pieces then. If we compose the test set from different subscans, does it solve the problem? Not at all. For example, we have a number of aerial scans. The scans can be retrieved from different heights, different scanners, in different weather. So, if we add the pieces of a single scan both to the test set and to the train set, we will get a non-challenging test again. The rule is: the pieces of a single scan may not persist both in the test set and the train set, if we want to train the classifier once for the whole domain. Do the test set and the train set come from the single distribution? No! But we need to neglect the theory in favour of practice.

One could say that it is reasonable to use cross-validation here. Well, it makes a sense. According to Wikipedia, there are three types of cross-validation:

• Repeated random sub-sampling validation
• K-fold cross-validation
• Leave-one-out cross-validation

According to the rule, only k-fold x-validation can be used, and each fold should contain points from its own scans. But it is very laborious to label scans. It takes more than 20 hours to label a standard million-points scan. So, we cannot have a lot of scans labelled.

This is not the only problem with testing point classification. Since the single point does not tell us anything, we should consider some neighbourhood of it, and approximate it with a surface. For every point in both sets there should be some neighbourhood. The problem is solved too you put the whole scan to the set.

future makeover...

i want to add some charm to our front porch....the first thing that would make it cuter would be wonderful ferns and great potted plants by the door, but considering it's january, i am starting by thinking about the front door. I have been in love with the front door from martha stewart forever. Our door is red and I plan on keeping it that way but I thought it would be fun and inexpensive to paint the No. 1 (of course i would add our correct house number) onto it. I also love the plaid doormat that was featured in country living in dec. (unfortunately it's on backorder). and the next thing would be to change the lighting the builders installed when the house was built. i am thinking something from barn light electric...i love the atomic topless guard sconce. so that's my daydreaming...i'll keep you up-to-date on what happens :)

*images courtesy of my pics, martha stewart, country living, coco doormats, barn light electric

Paleo is Going Mainstream

There was an article on the modern "Paleolithic" lifestyle in the New York Times today. I thought it was a pretty fair treatment of the subject, although it did paint it as more macho and carnivorous than it needs to be. It features three attractive NY cave people. It appeared in the styles section here. Paleo is going mainstream. I expect media health authorities to start getting defensive about it any minute now.

[2013 update.  Did I call it or what??]

baby it's cold outside...

it's suppose to be super cold this weekend so here are some things to ponder if your bored:

1. have you seen 500 days of summer? soooo good and the guy is really cute in it...you should really watch it and then google and see where you recognize the guy from (it will make you laugh).

2. feeling like cleaning? i have a huge obsession with mrs. meyers (kind of funny since i am not good at cleaning)...geranium and basil are the only ways to go. i can't live without the hand soap, counter top spray, and all purpose cleaner (money saving tip: buy the all purpose cleaner and refill your counter top spray bottle will a little of that and warm water). i have been known to say 'if no one was looking i might just spray this stuff and wear it as perfume'....

3. i heart dunkin' donuts coffee. i told warren that i think that i could be one of those people who could have the bumper sticker (i know, can you imagine)! coffee snobs are just dying at this revelation! the best is getting it fresh from the drive thru and having them put cream and sugar in it for you....yum.

4. feel like venturing out in the cold....tj maxx has designer's guild sheets and shams...really cute...really cheap (like $3.99 cheap). i got sheets and shams for taylor's room...they are getting monogrammed as we speak!

5. if you have kids, i am obsessed w/ mini boden and they are having a huge sale right now. i am bummed that this isn't available in taylor's size but they have TONS of cute things that i could totally buy.

okay, i'm off to charlotte to battle the cold and hopefully make a stop at south park and phoebe's new store! see you next week...stay warm!

*image courtesy of what anna loves

my color or lack of...

lots of people have been asking to see pictures of the house...believe me, you will definitely see pictures, as well as projects that i hope to be starting soon...but right now, i feel like it's lacking my favorite color and that i am swimming in a sea of khaki....i like khaki, don't get me wrong, i just like it combined with other colors!

*images courtesy of creature comforts, massie color, sweetie pie, domestikated life

you know, i'm not one to stick to a resolution....and now that i think about it, i don't think i really make a point of making them....does that make me a terrible person?!! but i will say that when i saw that 'love life' pic, i immediately thought 'that's pretty cool'- a great motto to live by....

*images courtesy of copy cat chic, harper and harley, high street market, lotus bleu