Thursday, June 16, 2016

Deep learning

A latest fad is deep learning which I characterize thus: sophisticated manifestation of fuzzy approaches abetted by numeric modeling and by Monte Carlo walkabouts. It is very expensive.

First some pointers to material:
Now, as a way of introduction, this approach supported AlphaGo's success. That is, a game that was supposed to be hard for computers was not. Too, people are lining up for jobs.

You know what? Many of these jobs are to pick people's pockets via the commercialism of the web/cloud.

Some think that they can get into this through a boot camp. But, realize, it takes a PhD level of work in order to know the guts. And, that takes time and effort.

Personally, for me, I am thankful. Finally, I see the world gaga about that which is the core of Truth Engineering's motivation.  

Remarks: Modified: 06/16/2016

06/16/2016 --

Tuesday, May 31, 2016

Content and context

Context: OfficeLiveContent, and all that.

This topic comes up again with my experimenting with an install of Windows 10. After going through the work and wait, I went into my personalization mode. So, then things looked like they did on Windows 7.

That was about a day. Of course, I didn't have to watch. Too, I was in and out, so it was not solid time.

Then, I tried to get to work. But, the thing was running hot. Let's say, enough to get my attention.

On a closer look, memory was about full. The CPU was maxed out. What? Well, the biggest culprit for the heat was was the memory manager that was compressing. After I got rid of some dinosaurs not being used, memory went down to about 50% usage (I have 6G).

But, the processors were at 100% (something chasing its damned tail). And, I have a dual set that was the latest four years ago.

All I was doing was typing a post like now. I multi-task, but I know how to balance things. I do not need the supposed smarts which are not very insightful.

Let this old guy tell how.

Anyway, I went back to Windows 7 until I can work a strategy to get the compression removed. If that is not possible, I'm stuck. Android? Linux?

I have been using this beast for four years now. Publishing. Computing. In my way which is not trivial. I know what the hell that I am doing. BTW, I had a BSOD the other day but now what was going on. Altogether, there have been only a handful over the past four years (as opposed to Vista/XP that seemed to like to barf). Too, Windows 7 has so many updates/installs accumulated that the tangle is a nightmare.

Finally, to the theme. We use computers to manage content within a context. Some of these are very much non-trivial. I'm not gaming. I'm not TV watching. I deal with intellectual items and use the computer as an assistant.

I do not need Cortara (whatever).

Listen up, folks. Of my generation and the leading edge of the boomers, we have people that know computing. They know what they want to do. We need to be able to configure a system that allows this type of work. Keep the flim-flam out.

If I want to do that, I'll get another system just for playing. Four years ago, it was the move from OfficeLive that opened up doors. Now, it's this. Fun times.

Remarks: Modified: 05/31/2016

05/31/2016 --

Tuesday, April 19, 2016

References and more

As the world moves, so does the universe of knowledge. So, one needs to keep an active eye. One way that we will do this is via collections of sites to watch.
Knowledge, nowadays, involves the cloud which entails the computational which is dependent upon mathematical insights. In short, STEM as the new type of sand (into which heads are thrust - not want to know type of thing).

Remarks: Modified: 05/31/2016

04/19/2016 --

Sunday, April 17, 2016

Mania or hysterics

A few days ago, we did a quick summary, related to intelligence which is a poorly understood thing, to say the least. Truth engineering deals directly with this subject.

Some have touted artificial intelligence as a means to understand ourselves. At least, again, some part of ourselves. And, we have no issue with a lot of this. In fact, science needs this thrust.

But, artificial intelligence? Actually, we know "artificial" as it pertains to the mode and means, which is via computing (various ways). And, with regard to that, we have a whole lot to say.

Right now, though, we can see two types of reaction that are getting press. Say, I saw this being discussed on a national TV show (Sunday morning variety) with youngsters weighing in with gravitas. Well, it's nice to hear certitude being expressed. Except, and a big except, science is provisional.

Is that being lost in all of the mania and hysteria? What? The mania deals with strong AI (as in, suggesting that the artificial will trump humankind in significant ways - let's talk limits, there). Hysteria? At the same time, we have those saying things like "summoning the demons" as the reality in this situation.

Now the hysteria can be thought of as a paranoid mania. So, we are talking sides of the same coin.

Per usual, there are those in the middle where the real rational minds venture, in just about anything.

Given our technical focus, we can look at limits on technology and natural constraints on modeling as important items to know more about. Too, there are capabilities that humans can exhibit if given the chance. The main problem is that the quantifiers have had a field day.

In some cases, they have accumulated almost uncountable amount of wealth. Yet, to what end? And, how is it that we need such for determining truth?

Remarks: Modified: 04/17/2016

04/17/2016 --

Wednesday, April 13, 2016

Boomer lessons

This post continues the look at Beans and such. In short, we are all of a cohort. By collection and analysis, we (the collective) have sufficient data that can be reasonably useful.

So, we can produce a chart like this that shows an expectation for the number of years one might plan for. Of course, we could all plan to live to 150. Would that make sense?

The top chart is for the early boomers. Then, we look at the situation during age 75 to age 85. Taking the top one first, notice that the 30-year view vanishes about the age of 70. Essentially, some will live to 100 and beyond. But, based upon data-to-date,  not many.

5 year look? It is quite rational. In fact, in the below, we would expect that one would plan for five years at a time. Then, the plan would be updated yearly.

---

Notice, that the green line runs down quickly. It, basically, says that, after age of 83, one ought not put much thought into planning for 20 years. Now, that does not say take no thought. As the line does not go to zero (see below). The red line is for 10 years. At 93, there is some noticeable chance of going for 10 years. But, at 93, the likelihood of another 5 years is more than one might imagine.

Aside: The IRS gives one a couple of years, each year, as one goes beyond 100. Also, there is this adage: live like there were no tomorrow; plan as if there were to be many.

And so, the issue of outliving your money is not to be taken lightly when planning. For this, we have all sorts of things, like long-term care insurance, etc., to think about. However, note, too, that some of these insurance companies are now telling their customers to not expect the agreed-upon payout and to belly up to the bar with a higher premium payment. That is, many of these contracts (some from years ago) are being re-written.

---

Now, we have some expectation of duration that can quantified. But, what about the flow? Well, first, let's consider that there is income and expenses. For income, many of the advisers like to push people toward the higher risk types of things. Sheesh. Sacrificial lambs is how we ought to see these older victims.

The top chart shows (bars) how a constant return can be used for planning. Now, lots of people are going to try to argue that the higher the return, the more you can spend. True. But, the higher the return, the higher the risk.

Planning ought to be conservative. Then, if there is gravy (home runs), great. However, if done right, this type of planning is an almost perfect hedge. As one also needs to consider the time involved. Of course, with a short time frame (and spending down to the last penny), one can be extravagant. But, we are talking flow over the longer haul.
In and out
65 and 75

Flow? One thing that anyone of normal situations ought to do is have a budget that looks at income and expenses. If not that, at least, have some idea of where the money comes from and goes (and whether one is getting further into debt).

There was a thing called zero-based budgeting that was the rage for awhile. It can work for those who are planning their future, at any age. The nearer one is to retirement, the more one ought to know in this regard.

Aside: I have 15 years of spreadsheet (several sheets per year) details, by category, that started prior to retirement. This is very easy to do and can be enlightening. For those who travel, a few months of daily doing the equivalent of an expense report for a trip ought to be easy. If you are traveling, there are extra expenses; however, some of those at home continue.

In the top chart, we are looking at the inflow and outgo (as a function of number of years expected). "dly" means that the flow was normalized to a day. This type of discretization is not that hard to do. And, it is very useful when you are looking at any type of outflow in relation to income or just in its own right. Sometimes, outflow can get heavier than normal.

Of course, "#yrs" relates to the possible number of years (with probability) that one can expect.

What the bars indicate is that the number of years does influence flow. If you want to spend your money with no control, then, expect it to not last (or look for bailouts).

BTW, there is a scale on the flow here that is immaterial. The notion in this case is comparative looks at the future states. However, does this imply that there would be a residue left for those who may or may not misuse the gift? Sure. That is another subject to address.

And, there is only one rule to success. Spend less than income (over all, the ins ought to outweigh the outs). If one cannot do so now, one ought to try to plan to get to that state of affairs. In this case, the out-flow is actually less than the 30 year figure (net savings situation).

---

So, expenses (and incomea) ought to be reduced to a daily average. With this view, one can relate the outs and ins so as to have some notion of whether one is bleeding or not. That is, when outs are high, they can be normalized to something that is consistently comparable.

For times of heavier outflow, there are considerations of category that come into play. For instance, staying in a super expensive hotel for 10 days averages out to less than the per night charge over a 30-day period. Trick? No, how one trickles out from one's bucket (and income) is a matter of choices. Too, major things (appliances, vehicles, etc.) would be handled in a manner so as to recognize the service life (etc.).

By doing this type of exercise, one can look at flows (out and in) by category. Which, then, gets back to some thoughts on inflow. Of course, one might like (or think that they might want) to hit a home run (no envy of Zuck or the like - see near-zero, below). But, think of the risk. Even the young ought not be too loose with risk, if they are thinking of the future. I know, the horizon can be a long way off depending upon age. We'll get back to that in future posts.

Right now, the older folks ought to think of near-zero. That is how our economy runs. That is, for those who make it big, legions (very many) lose (lose and lose and ...). Thanks to John Nash, et al, we have a very warped view of reality that has crept into acceptance. And, near-zero suggests, strongly, that the older folk be wise and not let old (or young) advisers get them into trouble. Those advisers will not bail the old folk out when things collapse.

Sheesh. Just read today of a mania. ETFs related to leveraging against the Janets of the world. Does the craziness never cease?

Remarks: Modified: 04/13/2016

04/13/2016 -- We also covered beans in the oops sense.




Monday, April 11, 2016

Beans and such

Back in 2009 (we started this in 2007, right before the display of idiocy arose everywhere, densely), we wrote about beans (as in, our  our daily bread). So, let's do this, again, and again.

I am 11 years retired, now. So, I was able to watch all of this silliness, from afar and up close (being slapped silly by bankers who tell me that they do not want, nor do they need, deposits - say what?). Too, my whole life has been analytically oriented, albeit in an unorthodox manner (which is more boon than bane). Too, I know economics and that whole little bit of gaming, more than I like to admit. As, I am more interested in issues related to a sustainable economy than the mess (ca'pital'sino) that we currently have.

How to do the cleanup? Well, it is complicated. Okay? For those who might want to pay attention, though, there are some things that ought to be considered and understood.

And, as for going to an financial adviser, even if the such a person thinks that they are doing their fiduciary duty (what does this mean?), one might be better off thinking for oneself. But, where does one go for information that is not so embedded in the capitalistic chimera as to stink?

And, such information ought to address all ages. For now, let's talk about the later years, say like the boomers are facing now.

---

When I retired 11 years ago, the first thing I did was to look at morbidity and mortality (see below). One motivation was that there had already been deaths in the family. Too, I was on a plane and found a report (MMWR) left by some earlier passenger that I took to be a reminder.

After looking at M&M, I went into a deep dive on finances and then intelligence. After that, there were many more topics covered. This post and the next are looking a the first two.

---

If we are to talk beans, time has to be considered. When we are young, our life has a long horizon, very long for some people. Also, the earlier years are usually spent working. At some point, the later years, and how they will be funded, come up. For the younger set, the earlier that you think of this, the better, as when one does get to a certain age, there is a quicker pace of departures from a cohort.
CDC of HSS 

The table shows expectancy for white males in the U.S. in 10-year increments. The last year added, 2003, is on the bottom. Notice the comparison with earlier years, especially those awhile ago. Life expectancy has been increasing. In fact, many of a cohort will live long past the average which is shown here (next graphic).

As an aside, the financial adviser is mentioned above. Many times those types were more into picking pockets than not, unfortunately. It seems that the whole financial industry is a big game, many times. We have gone into that and will again.

--

This graph shows a typical cohort's timeline in the sense of how many more years they can expect to be alive. Cohort? Think generation, but it is really those who were born the same year. Notice, the bottom axis is "Age." Why does it start at 20? Well, before that all of these lines were too close together to see any one. On the left, the value is the chance of living a certain number of years.

Slippery slopes
2004 data, U.S.
The curves are lookaheads from Age: 5, 10, 15, 20, and 25 years. As in, say Age of 55, the green line shows that over 90% of the cohort can expect to live another 5 years. 10, 15, and 20 years are a little less.

But, 25 years has a value of 80%. That is, 20% of those who are 55 years old can expect to be dead by 80. You see, that is higher than the average life expectancy. Also, the green line is still at 40% for a 5 year look-ahead at 90 years of age. What does that mean. At 90, the red and yellow lines are gone. At that point, one has a slight chance of living 15 years and a little more for 10 years.

IRS gives only a couple of years even when one is over 100.

The numbers come from tracking done every year. We know the size and characteristics of a cohort due to records. When people die, we know their cohort. After a number of years, we would know the cohort's remaining size.

This type of thing is part of an on-going statistical effort that relates to planning and management by the U.S. government. Below, we get into M&M, a little. As, one responsibility of the CDC is epidemiology, especially for those things that contribute to bad health and early death.

---

2004 data, U.K.
One thing of importance is that the health professionals have been recording deaths and causes, for ages. By now, we have a fairly good collection as each year the cohort database is updated. Of course, all of the time, new causes come about, except the broad categories do cover a wide range.

This chart is from U.K. data and relates two things. On the left, there is a scale of those who are still alive. As you would expect, this curve (pink) goes down. The other curve (blue) rises and does so fairly rapidly.

The range of Age here is from 60 to 85, so we are talking the early part of the boomers. As we mentioned (above), for the youngsters, these lines would be mostly straight. It is in the later times that the choices related to this come more to fore. And, that is the theme of these posts.

---

Finally, the look at M&M would not be complete to look at a typical year. As we have heard and seen, heart problems and cancer are the largest factors in early death. Notice that there is a huge bucket of "Other." As with any data report, we deal with definitions and outliers from that.

Morbidity and mortality report

---

Given that a good percentage of a cohort can go way beyond the average life expectancy, the issue of outliving one's assets is not to be taken lightly when planning.

And, planning is a continuous activity. Two main factors would be income and expenses. Next time (Boomer lessons), let's look at a good way to handle this. That is, plan in 5-year increments (updated yearly) with a daily focus for accounting.

Remarks: Modified: 04/13/2016

04/13/2016 -- We also covered beans in the oops sense.



Saturday, March 19, 2016

Truth and intelligence

There is a whole lot to say here, but let's just do a list of things to look at further, later.
  • I have been looking at Quora. It is more than interesting; as such, what is the appeal?
  • One thing that Quora does is allow questions to which people provide answers. There have been several themes that I noticed. One of these deals with IQ. For instance, people would ask whether the main rulers of the earth/world, at some future point, would have an IQ of 1000 or something or other. You see, numerics have warped our selves. 
  • In relation to queries like that one, people discuss Gardner's Multiple Intelligence proposal and such. Too, there are discussion about brain versus heart. For instances, one smarty was so because of an abusive environment. Later, the person realized, via a mentor, that such braininess is not properly balanced. BTW, Carl G. Jung is discussed, too. 
  • In short, IQ is one type of intelligence (see Gardner) for which there are tests. Some score highly where some means few. Most do not. Yet, a lot of these tests seem to have a common feel. So, one ought to be able to practice and score better, if one is able. But, the best of the tests are not timed. See this Quora answer for links. One test site: Dr. Jason Betts. That is nice, as it removes performance anxiety, silly dancing around, and competitiveness.  
  • Now, Jung was mentioned above. His thoughts about phenomenal issues that he addressed via Synchronicity are apropos to this discussion. In fact, Jung's work (or being reminded of such) gives me some hope for our future where the technologists would have run amok with their toys thereby, essentially, entrapping the rest of us (that is, human entanglement). But, much more on this.  
  • So, finally, to truth and intelligence. The truth bit, for me, comes down to truth engines and their application as supportive devices. Now, as I mentioned, technology seems to have run to where some brains (like Hawking) are bemoaning the end of the human race. Well, not. Those boys (for the most part, pale in face) will be reined in. Too, we need to have better human and machine cooperation.
  • You see, computation has holes in it. Despite the years of effort at error correction and the introduction of stability and a whole lot of other things, these holes are there; undecidability (in several senses) is a key issue. So, that theme will be addressed, time and again. 
  • But, intelligence? Yes, Google's system beat a Go player using what they term AI. Nice. Yet, not nice. However, this may very well be an example of technology being used without insight. We'll have to weigh in there. Though, I heard that Google has hired some philosophers in order to get it straight.
  • ... 
There will be much more to the topic. That is, though we have been at this for many years, we have just barely scratched the surface, so to speak.

Remarks: Modified: 03/19/2016

03/19/2016 --