Whoo hoo! I did it! An envigorating sense of accomplishment suddenly washed over me as the popup message announced that I completed the course and with a 95.3% average!
Honestly it felt a lot like topping out on that 5.9 climbing route (say what you like) you've been projecting for 6 months.
In my present job I need to provide accurate technical advice to researchers concerning the best computational resources for various research problems. Parallel processing architectures are necessary to achieve the exponential performance gains of the past and those gains are necessary to solve bigger and more complex problems. To use parallel architectures one must employ efficient, parallel algorithms many of which are used in machine learning. These factors (and a perverse, philosophical interest in AI) motivated me to take Andrew Ng's much touted online Machine Learning
course.
I learned how supervised and unsupervised
learning algorithms work, wrote efficient, vectorized code to implement these algorithms AND brushed up
on linear algebra and partial differential equations!
Many other courses are offered for FREE on Coursera. Sure, you have to pay the big bucks (and spend years) to get that coveted Stanford degree but you can learn what you need to know about all sorts of knowledge domains from the top experts in their fields for FREE. Now, what do you need to learn?
P.S.
did I mention that it was FREE?
Full disclosure - I did pay 79 bucks for this certificate I can hang on my wall, easily the best $79 I ever spent on college education!
Softmax cost function equation courtesy of Stanford's UFLDL web site
EDUberGeek
reflections on the latest issues and trends in technology and higher education.
Sunday, July 24, 2016
Friday, December 27, 2013
Digital Security - A Worthy New Year's Resolution
If you're looking for a worthwhile New Year's resolution, why not resolve to bump up your digital security a notch by changing all your passwords to new, stronger ones and resolve to never write them down!
As an IT practitioner I deal with passwords all the time. Between work and home I regularly use hundreds of accounts, each secured by a password. I don't EVER write them down! When it comes to password security there are just 2 rules to remember:
Rule 1: Length matters.
Rule 2: There is no other rule.
Lots of people use passwords like this one: Boston14. I picked 14 because it's almost 2014. I don't know why I picked Boston. This password meets many so-called strong password requirements being 8 in length and composed of at least one upper, one lower and one non-alphabetic character. Guess what, it's not a strong password. Neither is Yeller01 or SoccerM0M. Are your passwords kind of like these?
Here's how I construct secure passwords. First, I generate a very long password phrase that only I would know but that is easy for me to remember. String together several words that have nothing to do with each other but that you can easily remember. You can use numbers too but remember, it's the length that matters! For instance, I might combine a favorite writer, "Clancy" with a favorite Starbucks beverage, "Mocha" and a favorite band, "Boston". That gives me a password that is 17 characters long. That's a brute force search space of about 150 octillion unique combinations of upper and lower case letters. Of course, using a dictionary attack reduces the search space to a theoretical 100 quadrillion unique combinations (of any 3 of the 470,000 words in Merriam-Webster).
I recommend you read this excellent article on password security to better understand why length is the main thing to consider when creating strong passwords. I used the "Passfault" hack time calculator (read the article!) to evaluate my example passwords. The results:
OK, so what? So, I use this special long password to secure a password valet app such as:
KeePass Password Safe (my personal favorite), Dashlane or Password Wallet.
These apps keep your passwords securely organized and easily accessible from each of your devices. Basically you only have to remember the one "special" long password. All the others you can look up inside the app if you forget them. Perhaps the best part of this approach is that it forces me to record every account I have that made me enter a password. So, if some site I use gets hacked I just open up my password app, find the web site url, log in (it stores my username too!) and change my password to a new one.
Now go secure your digital world in 2014 and have a Happy New Year!
As an IT practitioner I deal with passwords all the time. Between work and home I regularly use hundreds of accounts, each secured by a password. I don't EVER write them down! When it comes to password security there are just 2 rules to remember:
Rule 1: Length matters.
Rule 2: There is no other rule.
Lots of people use passwords like this one: Boston14. I picked 14 because it's almost 2014. I don't know why I picked Boston. This password meets many so-called strong password requirements being 8 in length and composed of at least one upper, one lower and one non-alphabetic character. Guess what, it's not a strong password. Neither is Yeller01 or SoccerM0M. Are your passwords kind of like these?
Here's how I construct secure passwords. First, I generate a very long password phrase that only I would know but that is easy for me to remember. String together several words that have nothing to do with each other but that you can easily remember. You can use numbers too but remember, it's the length that matters! For instance, I might combine a favorite writer, "Clancy" with a favorite Starbucks beverage, "Mocha" and a favorite band, "Boston". That gives me a password that is 17 characters long. That's a brute force search space of about 150 octillion unique combinations of upper and lower case letters. Of course, using a dictionary attack reduces the search space to a theoretical 100 quadrillion unique combinations (of any 3 of the 470,000 words in Merriam-Webster).
I recommend you read this excellent article on password security to better understand why length is the main thing to consider when creating strong passwords. I used the "Passfault" hack time calculator (read the article!) to evaluate my example passwords. The results:
Password | Time to Crack | Size of Search Space |
Boston14 | < 1 day | 3 million |
Yeller01 | < 1 day | 45 million |
SoccerM0m | < 1 day | 8 billion |
ClancyMochaBoston | 220 years | 7 quadrillion |
OK, so what? So, I use this special long password to secure a password valet app such as:
KeePass Password Safe (my personal favorite), Dashlane or Password Wallet.
These apps keep your passwords securely organized and easily accessible from each of your devices. Basically you only have to remember the one "special" long password. All the others you can look up inside the app if you forget them. Perhaps the best part of this approach is that it forces me to record every account I have that made me enter a password. So, if some site I use gets hacked I just open up my password app, find the web site url, log in (it stores my username too!) and change my password to a new one.
Now go secure your digital world in 2014 and have a Happy New Year!
Thursday, September 5, 2013
When Culture Trumps Technology
I was dumbfounded when my colleague "Sue" (not her real name) announced in a meeting that the decision had been made to use "Redhat", our department file share server, for the collaboration and document management needed for an important, new, company-wide, committee and subcommittees. While it was not mentioned in the meeting, I knew that this committee would eventually need to share its work with a group of outside reviewers.
I gulped and blurted out, "Did you think about using Google Drive?" What a stereotypical tech guy thing to do. Skip all the discussion, consideration, dialog and feel-good collaboration and go straight to the solution. I knew that Google Drive would meet all their needs and minimize the need for IT to set things up and keep them going. I scheduled a meeting in which I would review the many advantages of Google Drive. I dazzled them with my brilliance as I showed them example after example of how easy Drive is to use.
Fast forward one week. The CFO sends me an email. He just got out of a committe kickoff meeting where it was announced they would be using ... wait for it ... flash drives for document management and collaboration.
Head exploding. Can't think straight. What to do? How can this be happening? I just showed them all the amazing wonders and advantages of Drive. I guess my educational/training meeting was a complete failure. I had been planning a big Google Drive launch this fall with training and promotional messaging and the whole deal. Now a highly visible group of campus leaders have put the bus in reverse and stepped on the gas!
As I stewed and thought about all of this I realized that a much deeper problem exists that gave rise to this problem. That is the familiar old "silo" problem that so many organizations struggle with. I recently heard this referred to as "culture trumps technology". Rather than view the IT folks as subject matter experts in technology and go to them with a list of requirements, a group of well-meaning users "solves" the problem on their own. They wonder why on earth the IT guy wastes their time ranting about some new-fangled way of storing documents in the "cloud".
Sensing that there was something bad about the "Redhat" solution (why else would IT have raised red flags?), they went back to find a better solution, but again on their own.
So, at the end of the day, the "silo" culture takes over and prevents the application of the right technology to the problem at hand. Successful, modern organizations are good at the latter. Organizations that can't or won't use computers for what they are good at, and people for what they are good at, will languish in mediocrity.
Are you using your technology people as experts? IT should be your first stop when you have a project that involves technology. And don't tell them the solution you need (I need 10 flash drives), tell them what you are trying to do (I need to share documents among 10 committees and post the results to a web site). Then heed what they say!
Don't let (bad) culture trump (good) technology!
I gulped and blurted out, "Did you think about using Google Drive?" What a stereotypical tech guy thing to do. Skip all the discussion, consideration, dialog and feel-good collaboration and go straight to the solution. I knew that Google Drive would meet all their needs and minimize the need for IT to set things up and keep them going. I scheduled a meeting in which I would review the many advantages of Google Drive. I dazzled them with my brilliance as I showed them example after example of how easy Drive is to use.
Fast forward one week. The CFO sends me an email. He just got out of a committe kickoff meeting where it was announced they would be using ... wait for it ... flash drives for document management and collaboration.
Head exploding. Can't think straight. What to do? How can this be happening? I just showed them all the amazing wonders and advantages of Drive. I guess my educational/training meeting was a complete failure. I had been planning a big Google Drive launch this fall with training and promotional messaging and the whole deal. Now a highly visible group of campus leaders have put the bus in reverse and stepped on the gas!
As I stewed and thought about all of this I realized that a much deeper problem exists that gave rise to this problem. That is the familiar old "silo" problem that so many organizations struggle with. I recently heard this referred to as "culture trumps technology". Rather than view the IT folks as subject matter experts in technology and go to them with a list of requirements, a group of well-meaning users "solves" the problem on their own. They wonder why on earth the IT guy wastes their time ranting about some new-fangled way of storing documents in the "cloud".
Sensing that there was something bad about the "Redhat" solution (why else would IT have raised red flags?), they went back to find a better solution, but again on their own.
So, at the end of the day, the "silo" culture takes over and prevents the application of the right technology to the problem at hand. Successful, modern organizations are good at the latter. Organizations that can't or won't use computers for what they are good at, and people for what they are good at, will languish in mediocrity.
Are you using your technology people as experts? IT should be your first stop when you have a project that involves technology. And don't tell them the solution you need (I need 10 flash drives), tell them what you are trying to do (I need to share documents among 10 committees and post the results to a web site). Then heed what they say!
Don't let (bad) culture trump (good) technology!
Subscribe to:
Posts (Atom)