Having freed myself of an enormous burden, I decided to spend a good chunk of Friday either sleeping, or staring off into space. I slept in today, and started to feel a little more like myself. Though I have a significant workload ahead of me, as I allowed it to pile up while working on the evil senior project, I decided to watch movies all day. Well, I suppose it would be more accurate to say most of the day. I spent the morning getting my bearings back, and wasted time chatting online. It was about three or so before I left to start watching movies. The first two, the Ladykillers, and Eternal Sunshine of the Spotless Mind were not playing as the declining Grand Haven 9, so I had to trek on up to Muskegon to witness them at the significantly better facilities. After witnessing those two movies, I swung back into Grand Haven to catch a showing of Taking Lives.

Ladykillers was indeed an enjoyable film to watch. It was refreshing to see a level of dark comedy that has not been in the movies in recent years, as Hollywood has become increasing unoriginal. Actually, the entire entertainment industry is really falling through, perhaps one of the reasons they fight so hard to protect their “intellectual property,” because they are incapable of creating new ideas. Anyhow, that is a rant for another time. Anyhow, the Ladykillers is a good movie. It has good comedy, interesting camera work, decent writing, and wonderful characters. I highly recommend it if you are into that kind of thing.

Eternal Sunshine of the Spotless Mind was definitely the best movie that I have seen in a long time. I felt that not only was it written well, it was acted well. This is particularly important, because the of the of the film. For the actors to maintain the emotional connection relevant to the moment, and the ethereal nature of the dream/memory state was amazing. I also found it to be intellectually, and emotionally provocative. The story addressed many issues dealing with human nature, love and memory. I suppose that I may have been doubly influenced, since in high school I was really into girls like Clementine. All things considered I have to highly recommend this film. I enjoyed it so much that I am planning on going out and renting other movies that Kaufman has been attached to: Confessions of a Dangerous Mind, Adaptation, and Human Nature. I really like Being John Malkovich, and Eternal Sunshine, so hopefully I will enjoy these other fine titles.

Now I must deal with the ugly business of dealing with Taking Lives. I cannot really think of anything good to say about it. I mean, well Keifer was in it, but his role was pretty much superfluous. There really was not anything good in the movie though. I guess the idea was someone interesting. I mean, and epic serial killer who assumes the identities of those he slaughters, etc… But, damn, was it poorly executed. It has been a while since I have been genuinely bored in an action/thriller movie, but I was in this. I think I traced all over my cup holder with my finger trying to entertain myself. The entirely movie seemed to be a grand culmination of derivative scenes, and concepts from typically actions/thriller movies. The writing was unintelligent and uninspired. I could go on more, but I am trying to keep my posts down to one page. Lets just say, that all things considered this was a horrid movie, and I cannot recommend that anyone other than Wesley Vanlandschoot go see this movie. And, since he did, I see no point of keeping it in theaters any longer.

So, I worked on Senior Project until 2 AM Thursday Morning, then got up again at 6 AM to go work on it more. I began what I now know as: “the horribleness.” From 8 AM Thursday, until 6 AM Friday morning I was in EOS. Only taking restroom breaks, no meals, not sanity. Only madness. The entire process was entirely horrid. I met with my American teammates at 8 AM Thursday morning, as well as the Swedes via IRC. My fellow Americans ditched before 11 AM to get to work. So, it was just the Swedes and me for another 11 hours. We made a lot of progress for a few hours, then we started having problems that never should have been.

All things considered we dealt with a lot of crap that just wasted time. One of the Swedes dropped before John (one of the Americans) came back at 10 PM. At that point it was two Swedes, and two Americans. We pushed through, and accomplish most of things we wanted to do. By the time both of the Swedes dropped, roughly 2 AM Friday morning, we were up to debugging the AI, the final component of the project. John and I debugged it until about 5 AM, where we hit a wall, and were unable to make any meaningful progress. So we worked on presentation material, and what not until about 6 AM.

At this point we left to take a shower and get our heads together. I got back at 7:30 AM, and John a little after 8. The Swedes showed up about twenty minutes later. The last member of our American team showed up about ten minutes before the presentation.

So, we presented, I was far from eloquent, I dare say it was my worst speech in college. I actually said “uh-um” quite often, which is uncharacteristic for me. At best the presentation was poor. Then, it got worse when we dealt with the presentation. After all the freaking work I had put in, in the past 24 hours, we had as much work, as worked the previous day at 8 AM. If I wouldn’t have been so tired I would have been furious. The failure in our presentation stole any satisfaction that I would have taken in the project being over. Now, at least for the immediate future, I will be haunted by what should have been.

Variations and Comparisons

The chapter begins by talking about the various revisions that the particle swarm paradigm has under gone since its original inception in 1994. Then they go into discussion about the areas in which modification is still being tested in order to yield better performance, particularly parameter selection. The two primary parameters, that have the greatest influence on the result of a test are Vmax and ?. Then they go into discussion about the traditional methods for implementing these two parameters into the system.

The next area they discuss are the results that changes in Vmax yield. They show the typical sine wave natural the particle swarm search algorithm, then show how the algorithm is affected by no restriction, and how smaller and smaller values for Vmax effect on the search pattern.

They then begin talking about ?, which is considered the “acceleration constant.” They go into an explanation about how this value is important in determining the type of trajectory the particles travel. There is also extended discussion about how various values of the can dramatically alter the form of the sine wave form. After showing the dramatic influence of varying ?, they talk about the effect of these alterations. From there they go into discussion about controlling the value of ?, in order to control the “explosion” or rather to control the trajectories of the particles. When discussion ? control they first address Clerc’s constriction coefficient. Then, they begin talking about inertia weighting.

The next area of discussion is the effectiveness of the particle swarm algorithm that comes from the interactions of its particles with their neighbors. This leads into a discussion about neighborhood topology, and its relation to human social interaction. This section is the setup for the following, where they discuss the sociometrics of the particle swarm.

In their discussion about sociometrics they focus on two primary topologies: the circle topology, and the wheel topology. In a circle topology individuals interact with their k nearest neighbors. In wheel all information is filtered through a single individual (like a government organization). Then there is discussion about how these sociometrics are going to be used in experiments.

The next area they discuss before going into the experiments is substituting cluster centers for previous bests. Then they go into an algorithm for determining their cluster centers in particle swarms. Then, they begin explaining their design for the experiment.

In the first experiment they ran their conditions on the four test functions defined in the design (Sphere, Rosenbrock, Rastrigin, Griewank), and Shaffer’s f6. After explaining how the experiment was run they go into a technical explanation of the results. While cluster centers may result in better problem solutions in general, the improvement to be expected is moderated by whether the other condition is an individual’s best or a cluster center (depending on the function).

The second experiment dealt with determining whether or not clustering affects the length of time requires to reach a criterion. This was addressed since human cognition if processed faster for familiar stimuli than for unfamiliar ones. They go into an explanation of how the experiment was setup, and how it was run. Then, they again went into an explanation of the results. Ultimately, clustering did significantly affect time.

After the experiments they go into a discussion about the empirical evidence, and how it shows that substituting cluster centers for the individual’s previous best positions can optimize the average performance per a fixed number of iterations. They continue to talk about this in more detail, and determine that the performance of the cluster center is usually better than an average individual, but worse than the best group of three.

The next section deals with adding the principle of selection into particle swarms. The system is a little different than in GA’s, since each generation is the same size. They liken it to a 1 to 1 genetic solution that is not typically used in evolutionary computing, but it is very effective in the particle swarm.

From there, they go into comparing inertia weights and constriction factors. They then going into an explanation of how these systems can be implement, and a method for testing them. Then, they performed these tests and discussed the results.

They go into discussion about some various other observations, and improvements. This leads into a discussion of asymmetric initialization. As all of their previous experiments had been done where the optima was 0,0. However, that is rarely the case, and generally you don’t know the optima when you setup a simulation. So, the conducted a test with initial optima set at offsets. Then, they explained the results of these tests and made some general conclusions.

They finish off the section with an extended discussion about how exactly particle swarms fit into the world of evolutionary algorithms. It seems that the EA people have historically accepted them, though they technically are not. Through contrast and comparison they show that while particles swarms are not really evolutionary algorithms, they cannot really be entirely separated into something else either.

Weapon: Dragon Sword
Ninpo: Inferno

Since I have been receiving a bunch of hits from people looking for strategies to defeat the seven level boss, I figure I might as well tell you how I defeated it. First of all, I am not going to kid you, she is one tough %$#^@, but she is beatable. I would say she is easily the most difficult boss up to that point, and this is coming from someone that had to face the second boss 76 times (let that be a lesson to you, never run out of healing supplies before your get to the boss).

I have read various strategies put throughout the Internet on this one, they are the typical you have to be patient, do wall jumps, and all the standard garbage. I mean seriously, who honestly think that saying something that generic is going to help someone? If I hear one more lamer tell me to wall jump I might go on a rampage, but I digress…the really key to this battle it to never stop moving, and a decent amount of health, some devil elixirs, and a lot of luck.

Now, for those of you that do not already know, she will dodge pretty much everything you attempt to throw at here when she is walking, or flying. Everyone once and a while she lets her guard down, and that is when you have to strike. Unfortunately, there is no obvious way to know. I have heard people say that she can be attacked immediately after she performs one of her attacks, but all that I have found it that I die faster try to attack her that way. So, essentially what you have to do it pick at her until you find a kink in the armor, which usually equates to luck, I’m not going to pretend there are some magical skill that allows this to work.

Once you actually manage to deal her a decent blow (such as a A + Y attack, or a strike from Ninpo such as inferno) she will drop to a knee for a very brief period of time. From what I have seen the best thing to do, is the first time you get her down immediately do an inferno, then keep basting her with as many as you can before she escapes, in my case I did five, which took away roughly half of her power. Once you are out of magic, I have found that it usually works to do a three hit combo then a jump slice. If timed properly this will knock her back to the ground as she leaves.

Now, I’ll talk about the variety of attacks that I saw her use in the eight times I faced her:

Energy Spheres: from time to time when you are hanging back she will summon up three pink energy balls that she will hurl at you, the best defense against these I have found it to first run around her in a circular fashion, and if they start to get close to you block and roll in a circle, this works most of the time, but not always.

Column Throw: while flying around she will sometimes grab some columns for the mysterious extra-dimensional plane that bosses get that attacks from. Anyhow, when she is about to do this there is a flash of green magic, that is your queue to run in a circle, like hell. If you are “the man” you can manage to dodge closer to her in order to attempt and attack. This never really worked for me though. I noticed that this attack is usually coupled with the energy spheres.

Magic Grab: this is easily her most irritating attack. It seems that if you ever start attacking her hastily through the air you get greeted by one of these. It’s basically a nasty column that shoots up and grabs you in a bubble, then hastily beats the crap out of your for a bit. This typically takes away at least half of your power. I’ve found that if she does this once, she will usually drop into a cycle of repeated use, so hang back for a while and dodge her attacks before you try to take her out again.

Regular Grab: if you try to get close and block, she will do what pretty much everyone else does in the game: her throw move. Unfortunately, the only way to prevent this ass kicking is to keep your distance if you are not attacking.

Combo(s): it seemed to me that whenever you got close to her and let down your guard she immediately began a short combo against you. Your best bet is to block as soon as you can and get the hell out of dodge. Don’t bother trying to counter. Almost all the time she ends the attack by using a flip kick, so if you are brave you can try to wait for that, and attack her immediately afterward. I never really got that to work though.

Charge: this attack always seems to come out of nowhere. People say that it happens when you are far away from her, but I have had her do it to me at point blank range. If you can actually react to it fast enough, the best thing to do is a charge attack against her. The ironic thing is you’ll do a little bit of damage to her, but remain undamaged yourself.

Foot Stomp: perhaps the most annoying attack in her arsenal. This attack always seems to come out of no where, and I rarely had a chance to escape it, but if you have fast reflexes, see seems to pause (though that is hardly the right word to use) for a fraction of a moment before she does it. Good luck with that.

Anyhow, I hope that I have prepared you adequately for this most horrid boss battle. I suppose that it really just come down to the fact that you need to pick a pattern to attack on, and get really lucky. Sometimes see does not even do a given attack for an entire battle. So, good luck all who attempt this feat.

If this strategy worked for you, or didn’t, or you have your own, feel free to leave a comment informing me so.