When Indexing Equals Compression: Experiments With: Fill & Download for Free

The idea that people could get along without emotion is the "big lie." Emotions are our system of dealing with the unknown, our system of deciding to push harder, run away, or go for it. We have lots of them. They can be intensely contradictory but this all creates a means of processing data that has yet to be recreated by engineers.Basically, our emotions operate like a Prediction market. The bidders are our memories. And, the call to bid is queued by similarity of perceptions/situations to events or perceptions stored in our memory. The process of recalling is very fast because it essentially bypasses our cognitive thought process though the same mechanism that habits use.The Power of Habit: Why We Do What We Do in Life and Business: Charles DuhiggHowever, emotions are both the source and product of habituation. Emotions provide both the rewards and punishments that trigger, modify and suppress habits. Thus as a perception is being recorded into memory, along with its perceptual components (auditory, visual, kinesthetic, olfactory/gustatory) the emotional loading is recorded too.When this memory is "triggered" by an event or perception that is similar to the event or perception stored in memory, it is recalled. The closer the match, the stronger the "signal" relative to the initial recollection.This is extremely similar to but not the same as the concept of the "indexed" query. However, in terms of computing, the processes the brain uses to compress and access data are very similar to concepts being used today in databases. Much of our sleeping brain activity is given over to the process of compressing data, indexing it, and associating it with existing data via associating it with already existing objects and classes and only saving the "interesting" variations.Slightly off topic but neuroscience is slowly approaching this same conclusion from the biological side.Best detailed hypothesis I've seen of the function of sleep!Emotions are different than Appetites/Urges. Both of these classes of sensation give rise to interior perceptions of pleasure or pain.Appetites and Urges are the expression of the body's demand for sustenance, liquid, air, warmth/coolness(temperature moderation), skin integrity (itch/pain), urination, evacuation, and procreation. These are instinctual level responses that can override all rational thought in acute distress though in varying degrees. You can will yourself to starve to death, but you cannot hold your breath until you die. (These sensations can be grouped into distressed or sated with gorged/bloated/dizzy/crowded as possibility for the appetites - food, liquid, air and comfort.)(over-hydration distress is rare though with normal kidney function)Emotions are a class of sensations apart from these basic urges of life. They can be grouped into three broad classes based on degree of spontaneity. But, this classification is more of a gradient than from impulsive to disciplined responses with emotional responses being in the center of this continuum. These groups are Impulse, Emotion and Attitude. This continuum is related to the amount of damping or constraint that restrains the amplitude and speed of reaching that amplitude of the onset of an emotional response. More later.In a broad generality emotions can exist in one of three exclusive classes, Anxiety, Aggression or Attraction. This is the good old "fight or flight" but a third state "seek or approach" is also a possibility. John C. Loehlin identified this third state in attempting to develop simple model of behavior that was used to generate a computer model called Aldous. And, this is very much the case for human emotion. In fact, as humans we actually live in this state up to 80% of our waking days. Aggressions and Anxieties are relatively rare but this is just "relatively" as 20% is still hours of each day.Beyond this, Dr. Hunter B. Shirley, isolated four separate domains of motivational action that form the basis of human emotional reasoning. These domains, Affinity, Identity, Mobility and Diversity represent basic motivational goals of the organism. This system allows each potential action, opportunity or threat to be analyzed from each of these perspectives simultaneously (in parallel) and the result of this analysis is yielded as fast as our systems can pattern match the perceptual cue to the recollected experience or experience class. This recollection process is very much almost automatic and the result is that fast.This is a fact of human psychology that marketing experts are more than happy to use to get you to buy their products.Psychology Hacks to Boost your MarketingThe conscious mind can take 5 times as long to catch up to the "gut" feeling you already have, and very often, your "reasoning" is just attempting to justify that feeling, not to reign it in. Which is also why, folks can often be heard saying some really silly things, as they try to justify their strong feelings but cannot command the facts because the "reasoning" part was completely skipped.This is not to say that this response is not reasoning though. In fact, the emotional result is the consequence of the perception in question being pattern matched against all recollections from four competitive goal sets simultaneously. This is a wonder and an amazing facility.I am going to focus on the center ground of this set of feelings after just touching on the extremes. Impulses are the result of matches where few alternate or offsetting experiences are recalled and thus children's behavior is dominated by impulsive displays and actions. As experience builds, countering and offsetting models and experiences temper this initial "rush" to judgement so that the result is more considered, the intensity of the feeling is dampened and focused, this sort response is called emotion here. As, even more experience builds, one can develop highly disciplined responses or even jaded responses which are called attitude.In the emotional range of expression, each of the four goal sets or domains of reasoning can have one of three states, Aggression, Anxiety or Attraction. This results in 12 basic emotions that all other emotions we recognize are combinations, underdamped (impulses), critically damped (emotions) or overdamped (attitudes). Impulsive Aggression or Anxiety often is at very high magnitude, lower levels resolve into emotion typically but impulsive emotions that are tinged with anxiety are often called "slimy" and when tinged with aggression, "crude."Here is a chart to help you recognize each of these twelve basic emotions.This is a quick map of the twelve basic emotions and how these compound into the great global compound emotions of Love, Hate, and Grief.These basic emotions can form compounds of two or four emotions depending on motivational correlation, and these compound can be of equal or differing intensities.So just like a color wheel, this system spins out definitions of emotions and meaning as we consider mixed emotions in this model. A partially ambivalent emotion is one that combines and attraction with an aggression or anxiety. These are handy ways of softening otherwise outright or naked displays of anxiety or aggression. But, the human mind interprets this very well and the "catch" is telegraphed.So instead of Admiration = Pride + Affection you get things like Condescension = Affection + Contempt. And, thus like a decoder ring, this chart also allows you to deconstruct that display of Exasperation into Anger + Affection. And, then you know, they are frustrated about something, and you can focus on that by converting Anger into Desire.. and thus Exasperation into Hope.Beyond the partial ambivalent emotions are a set of fully mixed emotions or ambivalent emotions that mix aggression and anxiety in a couple. I will do global compounds and global ambivalences some other time.Fully ambivalent emotions are gut wrenching difficult experiences for the person experiencing them and for people coping with the contradictory messages they send. So rather than Sincerity = Pride + Curiosity, Hypocrisy = Shame + Disdain which is a false kind of sincerity. Snobbery is also another form of false sincerity. And, envy is a false form of determination.However, when you look at these as cues to motivation and also as the preparation for action, emotions literally are our means of interpreting our own control signals. They are a feedback system that tells the mind just what it is thinking and doing. They also provide reward and punishment for building habits and achieving goals. In short, without them, we would be lost.

I attended a Microsoft Offcampus drive for Intern Hiring in 2019 and here’s my Interview Experience.Round - 1: Online Coding Round on Mettl WebsiteMicrosoft Conducted an online test named TechSetGo on Mettl website.3 Coding questions are required to solve within 90 minutesEach problem is a functional problem.Given an array find the Majority element in the array (an element whose frequency in that array is strictly greater than half of the size of the array.) and if there is no majority return -1.Find the sum of the top K scores in a competition.There are N colours into which a chameleon can change into. There is a Matrix of size N x N where for each element i x j in the matrix if the element is equal to 1 then the Chameleon can change from colour i to colour j otherwise the Chameleon cannot change from colour i to colour j. Currently, the Chameleon is in colour X and it wishes to change into Colour Y. Find the minimum number of colours it needs to change to be in Color Y.Different people got different sets of problems but they were easy to solve.I was able to solve all 3 problems and was invited to an Interview in Hyderabad.Round - 2: Group Fly RoundThis Round is a pen-paper round where you are supposed to write the code on paper for the problem given.2 coding questions are supposed to be solved in 45 minutes.Given a Matrix of Non-negative Integers. Find a path with maximum sum, from top left (0, 0) to bottom right (N - 1, M - 1) and at any given point in the matrix, you could either move right(x, y + 1) or down (x + 1, y).Given a Binary Tree find the lowest common ancestor between 2 nodes N1, N2.Few Microsoft Employees were present to help you if you have any doubt regarding the questions.Round - 3: Technical InterviewFrom this round, every round will be a F2F round. Interviewer asked for my resume and asked me to introduce myself.Interviewer was really friendly and was focusing on how I am approaching the problem and how I was able to convert my thoughts into code.Why are you interested in CS when you are from ECE?Then he asked me few OS questions.Now he dived into coding questions.Given a 2D character grid and a word. Return true if the word is present in the grid otherwise return false. A word can be matched in all 8 directions. I first explained my backtracking approach and while writing the code I explained a few sections of my code to keep him in the loop. After I wrote the around 3 / 4 part of the code he said it is ok and now we can move on to the next problem.Given 2 Binary Search Trees merge them into a single unified sorted array. I explained my approach to him. I will create 2 arrays that contains the in-order traversal of the 2 trees (in order traversal of the binary search tree gives us the sorted version of that tree) and create a 3rd array where I will merge these 2 arrays like merge operation in merge sort. After I explained he asked me to code it and after going through my code the interviewer asked me if I could solve the problem using only 2 arrays instead of 3. After a couple of hints from the Interviewer, I was able to solve it. Interviewer asked me to generate/create some test cases for this problem.Given a Binary Tree write a code for the iterative preorder traversal. I was able to code it using stacks.Round - 4: Technical InterviewInterviewer first Introduced himself to me. He is a Senior Software Engineer in Microsoft and has been working in Microsoft for past 7 years. After going through my resume he asked me about one of my projects. I was able to give a good explanation of my project.Interviewer asked me a few questions on OOPS Concepts and Now he gave me a situation where he recorded an apple how an apple in fresh state decays over a period of 5 - 6 days and he wanted to show this some audience since the audience will not sit watch for 5 - 6 days how will I compress / shorten it to a 2 - 3 minute video. I told him that I will sample the video for 1 second at every 1 hour so that the entire video will be shortened to a 2 - 3 minutes and the significant changes in an apple are recorded.Suppose the initial video is of 5 days long 5 x 24 = 120 hours. I will take snapshots of the video of 1 second after each hour at 0hrs, 1hrs, 2hrs, 3hrs, ….., 120hrs. So overall there will be 121 seconds of video ie: 2minutes and 1 second.He asked me what are the significant changes in the apple during its decay process.I told him that there will be a change in colour and change in shape of apple during its decay. He was satisfied with my answer and then moved onto a coding question.I was asked only one coding question in this round.Question: Given 2 numbers in LinkedList form find the sum of two numbers and return the head of the LinkedList which contains the sum of these two numbers.1->2->3 + 4->5->6 = 5->7->9 I told him that we can do it either recursively or iteratively and explained him both of the approaches and he asked me to code in any one of the approaches. I choose the iterative approachIn this approach, I will create a new LinkedList called result and then I will reverse the two input Linkedlists and then move through each and every node and add both the node’s contents and the carry of the addition will be used for the next node’s addition. and finally, reverse the result LinkedList.Here you need to check that one number might have more digits than the other 6->7->8 + 2->3 = 7->0->1  You also have to check out for carry-out generated.9->9->9 + 1 = 1->0->0->0  I wrote the code for it and there I was using 3 while loopswhile (number1 != NULL and number2 != NULL) {  // Some operations }  while (number1 != NULL) {  // if number1 has more digits than number2 then residual operations will take place here. }  while (number2 != NULL) {  // if number2 has more digits than number1 then residual operations will take place here. } He asked if I could do it using only one loop and I made some tweaks to the code to get the result.while (number1 != NULL or number2 != NULL) {  if (number1 == NULL and number2 == NULL)  break;  else if (number1 != NULL and number2 != NULL) {  // some code goes here.  }  else if (number1 != NULL) {  // some code goes here.  }  else {  // some code goes here.  } } He was satisfied with it and asked if I can do it without calling the reverse method for the result LinkedList. I didn’t get the idea initially but he gave me a hint that “Do you know any data structure that follows LIFO Principle.” then I wrote the code for the problem using stack and he was satisfied with my code.Round - 5: Technical + HRI was interviewed by a Manager (don’t remember the team’s name). He told me that he is working with Microsoft for the past 18 years.Firstly he went through my resume and was interested to know why I was interested in CS when I am from the ECE. I have won some local coding competitions and he asked me about them how I learnt data structures and algorithms and how did I practice and what were the hurdles that I encountered while learning.He asked me a single coding question. Given an array find the longest maximum sum subarray if there are multiple answers go with subarray with the least starting index. I told him about the Kadane’s algorithm and wrote the code for it. He went through my code and told me that he was impressed with my variable naming. He was satisfied with my code and asked me if there are any questions from my side. I asked him a few questions like what kind of projects an intern will be assigned during his/her internship and what are the teams that are present in MS-IDC Hyderabad. I finally asked when will I get the result of the interview he told me that I will get it by Monday (Interviews were held on Saturday) and the result did come on Monday but only after 4 weeks :((.Result: Selected.PS:I am not quite sure about the duration of the technical interviews because I didn’t keep track of time but it was something like around 45 minutes - 1hour.Edit:I forgot to mention that the Interviews were easier than I expected it to be. Anyways I am sure I would have cleared them even if they were harder.

I wrote a post about exactly this some time ago on Stack Overflow: Page on stackoverflow.com. I also wrote one for a similar question, but about C# instead of Java, but many (most) of the same comments apply at least equally to Java: Page on stackoverflow.com.Those both attempt to narrow the problem to looking only at the language (and the "default" implementation--Sun's for Java, Microsoft's for C#) itself. There are quite a few other points that have substantial effects on speed though.Just for example, if you want to package some functions up for dynamic linking in C or C++, you normally use the native executable format for the target OS (e.g., ELF for Linux, PE for Windows). Since this is heavily used, the OS people have put immense amounts of work into optimizing it nearly to death and back.To get roughly the same thing with Java, you typically put the code into .jar files. A jar file is really just a zip file with a different name tacked on. It compresses and indexes the constituent files. When they were thinking in terms of Java applets being downloaded over relatively slow internet connections, that made a lot of sense--the time you spent de-compressing the file was much more than repaid in reduced download time.For code on a local file system, that doesn't necessarily work out quite as well though. It still takes time to decompress the code, but most of what you save is hard drive space that currently (August of 2015) costs around $40US/Terabyte (in other words: so cheap it doesn't matter).Worse, this can end up actually wasting much more expensive space in RAM. If you have two native processes that use the same DLL/.so file, the OS will normally just map the same physical memory to both processes. A heavily used library might easily be mapped simultaneously to dozens or even hundreds of processes. With jar files, each JVM ends up allocating its own space for the decompressed data (though multiple processes running in a single JVM can share common data).Most of these could be reversed. For example, Java's use of jar files is actually done via what's basically a plug-in. There's no particularly good reason Oracle couldn't put their code into something like an ELF or PE file, and provide a new plug-in that worked with them instead of zip files.Likewise, although the official documentation denies that Java has pointers, the reality is that Java has pointers practically everywhere. This can lead to substantially sum-optimal use of memory--especially cache. For example, consider something like:class foo {   bar b;  baz c;  qux d; }; Assuming `bar`, `baz` and `qux` are classes, Java and C++ will lay these out in memory quite differently from each other. In C++ you'll get a single contiguous chunk of memory that contains the instance data for `b`, `c` and `d`. In Java, `foo` will contain three pointers, and `b`, `c` and `d` will each be allocated separately on the heap. Given how Java's (usual) memory allocator/garbage collector works, there's a decent chance that they'll be pretty close to each other--but no guarantee of it (and accessing each will still involve going to the pointer in `foo`, and dereferencing it to get to the actual data.If the JVM can devirtualize all the function calls in all of those, it might also be able to allocate them as a single block of data--but experience shows that while it may theoretically be possible in some cases, it's pretty unusual to see it for any significant code in real use. At least in theory, a JVM implementer could work much harder at making this happen a lot more often.Exactly how much difference such changes would make would depend heavily upon what you do and how you do it. For a long-running server program that starts up, decompresses jar files, and runs for weeks or months afterwards, decompressing jar files is utterly irrelevant. Likewise, for cache usage it'll depend heavily on whether the extra pointers end up (for example) causing the cache to overflow so you use main memory a lot more. At least in my experience, it's a place that you see a huge difference between a benchmark and reality--and (as you'd probably expect) it's reality that gets the short end of the stick.