Search Results

Search found 463 results on 19 pages for 'train'.

facebook twitter digg google delicious technorati stumbleupon myspace wordpress linkedin gmail igoogle windows live tumblr viadeo yahoo buzz yahoo mail yahoo bookmarks favorites email print

Page 2/19 | < Previous Page | 1 2 3 4 5 6 7 8 9 10 11 12  | Next Page >

  • Need help with artificial neural network

    - by deckard cain
    I have an input data for neural network that consists of 2 vectors with 200 elements, that i got from some program for generating signals. So it is actually 2x200 input to my nnet. As target data, i have one 1x200 vector that i also got from the same program. That is my training data set. I gather as much of those sets as i want so i transfer them to matlab and save them as, for example, set1, set2,.... When i am creating neural net, using newfit function (backropagation algorithm and everyhting else is set by default because i am kind of unexperianced with neural nets so i will have to experiment) i'm creating it using set1 only for example. Then, when i am to train neural net i train it for set1 then load set2 and train for it and so on. so its like this function net = create_fit_net(inputs,targets) numHiddenNeurons = 20; net = newfit(inputs,targets,numHiddenNeurons); net=train(net,inputs,targets); load set2; net=train(net,inputs,targets); load set3; net=train(net,inputs,targets); load set4; net=train(net,inputs,targets); i am using 4 sets of data here and all sets have variables of same name and size. My quesion is, am i doing this the right way, because, when doing simulation in some other m file, i am getting bad results, and every time i get different results. Does it matter if i create network with one set and then train with others too, and does it matter what set do i use to train network 1st? Also, i am confused about the amount of neurons to use (im using in the example 20 but actually i tried 1, 10, 30, 50, 100 200 and even 300 and i get nothing). If you have any suggestions, i'd be glad to take them into consideration. Any help is welcome. thanks in advance

    Read the article

  • Concours de développement ce week-end dans un train pour la Méditerranée ! Départs de Paris, Nantes, Bordeaux... et Marseille

    BeMyApp organise un concours de développement mobile? dans un train ! Passez le prochain Week-end sur les bords de la méditerranée. Départs de Paris, Marseille, Nantes et Bordeaux Les participants devront développer des applications mobiles touristiques en utilisant les données mises à disposition par Bouches-du-Rhône Tourisme, le département de la Gironde et de la LoireAtlantique, la SNCF ainsi que les données nationales de data.gouv.fr La 17ème édition des WeekEnds BeMyApp sera unique en son genre car en plus de devoir créer une application en 48h, les participants devront programmer dans des trains en direction de Mars...

    Read the article

  • How to train users converting from PC to Mac/Apple at a small non profit?

    - by Everette Mills
    Background: I am part of a team that provides volunteer tech support to a local non profit. We are in the position to obtain a grant to update almost all of our computers (many of them 5 to 7 year old machines running XP), provide laptops for users that need them, etc. We are considering switching our users from PC (WinXP) to Macs. The technical aspects of switching will not be an issue for the team. We are in the process of planning data conversions, machine setup, server changes, etc regardless of whether we switch to Macs or much newer PCs. About 1/4 of the staff uses or has access to a Mac at home, these users already understand the basics of using the equipment. We have another set of (generally younger) users that are technically savvy and while slightly inconvenienced and slowed for a few days should be able to switch over quickly. Finally, several members of the staff are older and have many issues using there computers today. We think in the long run switching to Macs may provide a better user experience, fewer IT headaches, and more effective use of computers. The questions we have is what resources and training (webpages, Books, online training materials or online courses) do you recommend that we provide to users to enable the switchover to happen smoothly. Especially, with a focus on providing different levels of training and support to users with different skill levels. If you have done this in your own organization, what steps were successful, what areas were less successful?

    Read the article

  • Developer’s Life – Disaster Lessons – Notes from the Field #039

    - by Pinal Dave
    [Note from Pinal]: This is a 39th episode of Notes from the Field series. What is the best solution do you have when you encounter a disaster in your organization. Now many of you would answer that in this scenario you would have another standby machine or alternative which you will plug in. Now let me ask second question – What would you do if you as an individual faces disaster?  In this episode of the Notes from the Field series database expert Mike Walsh explains a very crucial issue we face in our career, which is not technical but more to relate to human nature. Read on this may be the best blog post you might read in recent times. Howdy! When it was my turn to share the Notes from the Field last time, I took a departure from my normal technical content to talk about Attitude and Communication.(http://blog.sqlauthority.com/2014/05/08/developers-life-attitude-and-communication-they-can-cause-problems-notes-from-the-field-027/) Pinal said it was a popular topic so I hope he won’t mind if I stick with Professional Development for another of my turns at sharing some information here. Like I said last time, the “soft skills” of the IT world are often just as important – sometimes more important – than the technical skills. As a consultant with Linchpin People – I see so many situations where the professional skills I’ve gained and use are more valuable to clients than knowing the best way to tune a query. Today I want to continue talking about professional development and tell you about the way I almost got myself hit by a train – and why that matters in our day jobs. Sometimes we can learn a lot from disasters. Whether we caused them or someone else did. If you are interested in learning about some of my observations in these lessons you can see more where I talk about lessons from disasters on my blog. For now, though, onto how I almost got my vehicle hit by a train… The Train Crash That Almost Was…. My family and I own a little schoolhouse building about a 10 mile drive away from our house. We use it as a free resource for families in the area that homeschool their children – so they can have some class space. I go up there a lot to check in on the property, to take care of the trash and to do work on the property. On the way there, there is a very small Stop Sign controlled railroad intersection. There is only two small freight trains a day passing there. Actually the same train, making a journey south and then back North. That’s it. This road is a small rural road, barely ever a second car driving in the neighborhood there when I am. The stop sign is pretty much there only for the train crossing. When we first bought the building, I was up there a lot doing renovations on the property. Being familiar with the area, I am also familiar with the train schedule and know the tracks are normally free of trains. So I developed a bad habit. You see, I’d approach the stop sign and slow down as I roll through it. Sometimes I’d do a quick look and come to an “almost” stop there but keep on going. I let my impatience and complacency take over. And that is because most of the time I was going there long after the train was done for the day or in between the runs. This habit became pretty well established after a couple years of driving the route. The behavior reinforced a bit by the success ratio. I saw others doing it as well from the neighborhood when I would happen to be there around the time another car was there. Well. You already know where this ends up by the title and backstory here. A few months ago I came to that little crossing, and I started to do the normal routine. I’d pretty much stopped looking in some respects because of the pattern I’d gotten into.  For some reason I looked and heard and saw the train slowly approaching and slammed on my brakes and stopped. It was an abrupt stop, and it was close. I probably would have made it okay, but I sat there thinking about lessons for IT professionals from the situation once I started breathing again and watched the cars loaded with sand and propane slowly labored down the tracks… Here are Those Lessons… It’s easy to get stuck into a routine – That isn’t always bad. Except when it’s a bad routine. Momentum and inertia are powerful. Once you have a habit and a routine developed – it’s really hard to break that. Make sure you are setting the right routines and habits TODAY. What almost dangerous things are you doing today? How are you almost messing up your production environment today? Stop doing that. Be Deliberate – (Even when you are the only one) – Like I said – a lot of people roll through that stop sign. Perhaps the neighbors or other drivers think “why is he fully stopping and looking… The train only comes two times a day!” – they can think that all they want. Through deliberate actions and forcing myself to pay attention, I will avoid that oops again. Slow down. Take a deep breath. Be Deliberate in your job. Pay attention to the small stuff and go out of your way to be careful. It will save you later. Be Observant – Keep your eyes open. By looking around, observing the situation and understanding what your servers, databases, users and vendors are doing – you’ll notice when something is out of place. But if you don’t know what is normal, if you don’t look to make sure nothing has changed – that train will come and get you. Where can you be more observant? What warning signs are you ignoring in your environment today? In the IT world – trains are everywhere. Projects move fast. Decisions happen fast. Problems turn from a warning sign to a disaster quickly. If you get stuck in a complacent pattern of “Everything is okay, it always has been and always will be” – that’s the time that you will most likely get stuck in a bad situation. Don’t let yourself get complacent, don’t let your team get complacent. That will lead to being proactive. And a proactive environment spends less money on consultants for troubleshooting problems you should have seen ahead of time. You can spend your money and IT budget on improving for your customers. If you want to get started with performance analytics and triage of virtualized SQL Servers with the help of experts, read more over at Fix Your SQL Server. Reference: Pinal Dave (http://blog.sqlauthority.com)Filed under: Notes from the Field, PostADay, SQL, SQL Authority, SQL Query, SQL Server, SQL Tips and Tricks, T SQL

    Read the article

  • Redirect network logs from syslog to another file

    - by w0rldart
    I keep logging way to much info (not needed, for now) in my syslog, and not daily or hourly... but instant. If I want to watch for something in my syslog I just can't because the network log keeps interfering. So, how can I redirect network logs to another file and/or stop logging it? Dec 10 17:01:33 user kernel: [ 8716.000587] MediaState is connected Dec 10 17:01:33 user kernel: [ 8716.000599] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:01:33 user kernel: [ 8716.000601] ==>rt_ioctl_giwfreq 11 Dec 10 17:01:33 user kernel: [ 8716.000612] rt28xx_get_wireless_stats ---> Dec 10 17:01:33 user kernel: [ 8716.000615] <--- rt28xx_get_wireless_stats Dec 10 17:01:39 user kernel: [ 8722.000714] MediaState is connected Dec 10 17:01:39 user kernel: [ 8722.000729] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:01:39 user kernel: [ 8722.000732] ==>rt_ioctl_giwfreq 11 Dec 10 17:01:39 user kernel: [ 8722.000747] rt28xx_get_wireless_stats ---> Dec 10 17:01:39 user kernel: [ 8722.000751] <--- rt28xx_get_wireless_stats Dec 10 17:01:44 user kernel: [ 8726.904025] QuickDRS: TxTotalCnt <= 15, train back to original rate Dec 10 17:01:45 user kernel: [ 8728.003138] MediaState is connected Dec 10 17:01:45 user kernel: [ 8728.003153] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:01:45 user kernel: [ 8728.003157] ==>rt_ioctl_giwfreq 11 Dec 10 17:01:45 user kernel: [ 8728.003171] rt28xx_get_wireless_stats ---> Dec 10 17:01:45 user kernel: [ 8728.003175] <--- rt28xx_get_wireless_stats Dec 10 17:01:51 user kernel: [ 8734.004066] MediaState is connected Dec 10 17:01:51 user kernel: [ 8734.004079] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:01:51 user kernel: [ 8734.004082] ==>rt_ioctl_giwfreq 11 Dec 10 17:01:51 user kernel: [ 8734.004096] rt28xx_get_wireless_stats ---> Dec 10 17:01:51 user kernel: [ 8734.004099] <--- rt28xx_get_wireless_stats Dec 10 17:01:57 user kernel: [ 8740.004108] MediaState is connected Dec 10 17:01:57 user kernel: [ 8740.004119] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:01:57 user kernel: [ 8740.004121] ==>rt_ioctl_giwfreq 11 Dec 10 17:01:57 user kernel: [ 8740.004132] rt28xx_get_wireless_stats ---> Dec 10 17:01:57 user kernel: [ 8740.004135] <--- rt28xx_get_wireless_stats Dec 10 17:01:57 user kernel: [ 8740.436021] QuickDRS: TxTotalCnt <= 15, train back to original rate Dec 10 17:02:03 user kernel: [ 8746.005280] MediaState is connected Dec 10 17:02:03 user kernel: [ 8746.005294] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:02:03 user kernel: [ 8746.005298] ==>rt_ioctl_giwfreq 11 Dec 10 17:02:03 user kernel: [ 8746.005312] rt28xx_get_wireless_stats ---> Dec 10 17:02:03 user kernel: [ 8746.005315] <--- rt28xx_get_wireless_stats Dec 10 17:02:09 user kernel: [ 8752.004790] MediaState is connected Dec 10 17:02:09 user kernel: [ 8752.004804] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:02:09 user kernel: [ 8752.004808] ==>rt_ioctl_giwfreq 11 Dec 10 17:02:09 user kernel: [ 8752.004821] rt28xx_get_wireless_stats ---> Dec 10 17:02:09 user kernel: [ 8752.004825] <--- rt28xx_get_wireless_stats Dec 10 17:02:15 user kernel: [ 8757.984031] QuickDRS: TxTotalCnt <= 15, train back to original rate Dec 10 17:02:15 user kernel: [ 8758.004078] MediaState is connected Dec 10 17:02:15 user kernel: [ 8758.004094] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:02:15 user kernel: [ 8758.004097] ==>rt_ioctl_giwfreq 11 Dec 10 17:02:15 user kernel: [ 8758.004112] rt28xx_get_wireless_stats ---> Dec 10 17:02:15 user kernel: [ 8758.004116] <--- rt28xx_get_wireless_stats Dec 10 17:02:16 user kernel: [ 8759.492017] QuickDRS: TxTotalCnt <= 15, train back to original rate Dec 10 17:02:19 user kernel: [ 8762.002179] SCANNING, suspend MSDU transmission ... Dec 10 17:02:19 user kernel: [ 8762.004291] MlmeScanReqAction -- Send PSM Data frame for off channel RM, SCAN_IN_PROGRESS=1! Dec 10 17:02:19 user kernel: [ 8762.025055] SYNC - BBP R4 to 20MHz.l Dec 10 17:02:19 user kernel: [ 8762.027249] RT35xx: SwitchChannel#1(RF=8, Pwr0=30, Pwr1=25, 2T), N=0xF1, K=0x02, R=0x02 Dec 10 17:02:19 user kernel: [ 8762.170206] RT35xx: SwitchChannel#2(RF=8, Pwr0=30, Pwr1=25, 2T), N=0xF1, K=0x07, R=0x02 Dec 10 17:02:19 user kernel: [ 8762.318211] RT35xx: SwitchChannel#3(RF=8, Pwr0=30, Pwr1=25, 2T), N=0xF2, K=0x02, R=0x02 Dec 10 17:02:19 user kernel: [ 8762.462269] RT35xx: SwitchChannel#4(RF=8, Pwr0=30, Pwr1=25, 2T), N=0xF2, K=0x07, R=0x02 Dec 10 17:02:19 user kernel: [ 8762.606229] RT35xx: SwitchChannel#5(RF=8, Pwr0=30, Pwr1=25, 2T), N=0xF3, K=0x02, R=0x02 Dec 10 17:02:19 user kernel: [ 8762.750202] RT35xx: SwitchChannel#6(RF=8, Pwr0=30, Pwr1=25, 2T), N=0xF3, K=0x07, R=0x02 Dec 10 17:02:20 user kernel: [ 8762.894217] RT35xx: SwitchChannel#7(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF4, K=0x02, R=0x02 Dec 10 17:02:20 user kernel: [ 8763.038202] RT35xx: SwitchChannel#11(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF6, K=0x02, R=0x02 Dec 10 17:02:20 user kernel: [ 8763.040194] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:20 user kernel: [ 8763.040199] BAR(1) : Tid (0) - 03a3:037e Dec 10 17:02:20 user kernel: [ 8763.040387] SYNC - End of SCAN, restore to channel 11, Total BSS[03] Dec 10 17:02:20 user kernel: [ 8763.040400] ScanNextChannel -- Send PSM Data frame Dec 10 17:02:20 user kernel: [ 8763.040402] bFastRoamingScan ~~~~~~~~~~~~~ Get back to send data ~~~~~~~~~~~~~ Dec 10 17:02:20 user kernel: [ 8763.040405] SCAN done, resume MSDU transmission ... Dec 10 17:02:20 user kernel: [ 8763.047022] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:20 user kernel: [ 8763.047026] BAR(1) : Tid (0) - 03a3:03a5 Dec 10 17:02:21 user kernel: [ 8763.898130] bImprovedScan ............. Resume for bImprovedScan, SCAN_PENDING .............. Dec 10 17:02:21 user kernel: [ 8763.898143] SCANNING, suspend MSDU transmission ... Dec 10 17:02:21 user kernel: [ 8763.900245] MlmeScanReqAction -- Send PSM Data frame for off channel RM, SCAN_IN_PROGRESS=1! Dec 10 17:02:21 user kernel: [ 8763.921144] SYNC - BBP R4 to 20MHz.l Dec 10 17:02:21 user kernel: [ 8763.923339] RT35xx: SwitchChannel#8(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF4, K=0x07, R=0x02 Dec 10 17:02:21 user kernel: [ 8763.996019] QuickDRS: TxTotalCnt <= 15, train back to original rate Dec 10 17:02:21 user kernel: [ 8764.066221] RT35xx: SwitchChannel#9(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF5, K=0x02, R=0x02 Dec 10 17:02:21 user kernel: [ 8764.210212] RT35xx: SwitchChannel#10(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF5, K=0x07, R=0x02 Dec 10 17:02:21 user kernel: [ 8764.215536] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:21 user kernel: [ 8764.215542] BAR(1) : Tid (0) - 0457:0452 Dec 10 17:02:21 user kernel: [ 8764.244000] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:21 user kernel: [ 8764.244004] BAR(1) : Tid (0) - 0459:0456 Dec 10 17:02:21 user kernel: [ 8764.253019] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:21 user kernel: [ 8764.253023] BAR(1) : Tid (0) - 045c:0458 Dec 10 17:02:21 user kernel: [ 8764.256677] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:21 user kernel: [ 8764.256681] BAR(1) : Tid (0) - 045c:045b Dec 10 17:02:21 user kernel: [ 8764.259785] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:21 user kernel: [ 8764.259788] BAR(1) : Tid (0) - 045d:045b Dec 10 17:02:21 user kernel: [ 8764.280467] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:21 user kernel: [ 8764.280471] BAR(1) : Tid (0) - 045f:045c Dec 10 17:02:21 user kernel: [ 8764.282189] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:21 user kernel: [ 8764.282192] BAR(1) : Tid (0) - 045f:045e Dec 10 17:02:21 user kernel: [ 8764.354204] RT35xx: SwitchChannel#11(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF6, K=0x02, R=0x02 Dec 10 17:02:21 user kernel: [ 8764.356408] ScanNextChannel():Send PWA NullData frame to notify the associated AP! Dec 10 17:02:21 user kernel: [ 8764.498202] RT35xx: SwitchChannel#12(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF6, K=0x07, R=0x02 Dec 10 17:02:21 user kernel: [ 8764.642210] RT35xx: SwitchChannel#13(RF=8, Pwr0=30, Pwr1=28, 2T), N=0xF7, K=0x02, R=0x02 Dec 10 17:02:22 user kernel: [ 8764.790229] RT35xx: SwitchChannel#14(RF=8, Pwr0=30, Pwr1=28, 2T), N=0xF8, K=0x04, R=0x02 Dec 10 17:02:22 user kernel: [ 8764.934238] RT35xx: SwitchChannel#11(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF6, K=0x02, R=0x02 Dec 10 17:02:22 user kernel: [ 8764.935243] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:22 user kernel: [ 8764.935249] BAR(1) : Tid (0) - 048e:0485 Dec 10 17:02:22 user kernel: [ 8764.936423] SYNC - End of SCAN, restore to channel 11, Total BSS[05] Dec 10 17:02:22 user kernel: [ 8764.936436] ScanNextChannel -- Send PSM Data frame Dec 10 17:02:22 user kernel: [ 8764.936440] SCAN done, resume MSDU transmission ... Dec 10 17:02:22 user kernel: [ 8764.940529] RT35xx: SwitchChannel#11(RF=8, Pwr0=29, Pwr1=26, 2T), N=0xF6, K=0x02, R=0x02 Dec 10 17:02:22 user kernel: [ 8764.942178] CntlEnqueueForRecv(): BAR-Wcid(1), Tid (0) Dec 10 17:02:22 user kernel: [ 8764.942182] BAR(1) : Tid (0) - 0493:048e Dec 10 17:02:22 user kernel: [ 8764.942715] CNTL - All roaming failed, restore to channel 11, Total BSS[05] Dec 10 17:02:22 user kernel: [ 8764.948016] MMCHK - No BEACON. restore R66 to the low bound(56) Dec 10 17:02:22 user kernel: [ 8764.948307] ===>rt_ioctl_giwscan. 5(5) BSS returned, data->length = 1111 Dec 10 17:02:23 user kernel: [ 8766.048073] QuickDRS: TxTotalCnt <= 15, train back to original rate Dec 10 17:02:23 user kernel: [ 8766.552034] QuickDRS: TxTotalCnt <= 15, train back to original rate Dec 10 17:02:27 user kernel: [ 8770.001180] MediaState is connected Dec 10 17:02:27 user kernel: [ 8770.001197] ==>rt_ioctl_giwmode(mode=2) Dec 10 17:02:27 user kernel: [ 8770.001201] ==>rt_ioctl_giwfreq 11 Dec 10 17:02:27 user kernel: [ 8770.001219] rt28xx_get_wireless_stats ---> Dec 10 17:02:27 user kernel: [ 8770.001223] <--- rt28xx_get_wireless_stats Dec 10 17:02:28 user kernel: [ 8771.564020] QuickDRS: TxTotalCnt <= 15, train back to original rate Dec 10 17:02:29 user kernel: [ 8772.064031] QuickDRS: TxTotalCnt <= 15, train back to original rate

    Read the article

  • Is it still too early to hop aboard the Python 3 train?

    - by Znarkus
    I'm still a beginner to Python, so I thought I could as well learn the newest iteration of Python. Especially since it is now 3.1 or 3.2 something. But it seems like many mayor modules are still only supported by 2.6. Like the python-mysql module; from what I read on http://mysql-python.blogspot.com/ it seems like 3.x support won't be seen in any near future. Do you use version 3, how do you get around these problems? Should I retreat to 2.6? If not, what should I use to connect to MySQL?

    Read the article

  • Optimizing Haskell code

    - by Masse
    I'm trying to learn Haskell and after an article in reddit about Markov text chains, I decided to implement Markov text generation first in Python and now in Haskell. However I noticed that my python implementation is way faster than the Haskell version, even Haskell is compiled to native code. I am wondering what I should do to make the Haskell code run faster and for now I believe it's so much slower because of using Data.Map instead of hashmaps, but I'm not sure I'll post the Python code and Haskell as well. With the same data, Python takes around 3 seconds and Haskell is closer to 16 seconds. It comes without saying that I'll take any constructive criticism :). import random import re import cPickle class Markov: def __init__(self, filenames): self.filenames = filenames self.cache = self.train(self.readfiles()) picklefd = open("dump", "w") cPickle.dump(self.cache, picklefd) picklefd.close() def train(self, text): splitted = re.findall(r"(\w+|[.!?',])", text) print "Total of %d splitted words" % (len(splitted)) cache = {} for i in xrange(len(splitted)-2): pair = (splitted[i], splitted[i+1]) followup = splitted[i+2] if pair in cache: if followup not in cache[pair]: cache[pair][followup] = 1 else: cache[pair][followup] += 1 else: cache[pair] = {followup: 1} return cache def readfiles(self): data = "" for filename in self.filenames: fd = open(filename) data += fd.read() fd.close() return data def concat(self, words): sentence = "" for word in words: if word in "'\",?!:;.": sentence = sentence[0:-1] + word + " " else: sentence += word + " " return sentence def pickword(self, words): temp = [(k, words[k]) for k in words] results = [] for (word, n) in temp: results.append(word) if n > 1: for i in xrange(n-1): results.append(word) return random.choice(results) def gentext(self, words): allwords = [k for k in self.cache] (first, second) = random.choice(filter(lambda (a,b): a.istitle(), [k for k in self.cache])) sentence = [first, second] while len(sentence) < words or sentence[-1] is not ".": current = (sentence[-2], sentence[-1]) if current in self.cache: followup = self.pickword(self.cache[current]) sentence.append(followup) else: print "Wasn't able to. Breaking" break print self.concat(sentence) Markov(["76.txt"]) -- module Markov ( train , fox ) where import Debug.Trace import qualified Data.Map as M import qualified System.Random as R import qualified Data.ByteString.Char8 as B type Database = M.Map (B.ByteString, B.ByteString) (M.Map B.ByteString Int) train :: [B.ByteString] -> Database train (x:y:[]) = M.empty train (x:y:z:xs) = let l = train (y:z:xs) in M.insertWith' (\new old -> M.insertWith' (+) z 1 old) (x, y) (M.singleton z 1) `seq` l main = do contents <- B.readFile "76.txt" print $ train $ B.words contents fox="The quick brown fox jumps over the brown fox who is slow jumps over the brown fox who is dead."

    Read the article

  • Le Web serait-il en danger face à la montée en puissance du mobile ? Pour l'écrivain et chercheur Danny Crichton, le « Web est en train de mourir »

    Le web serait-il en danger face à la montée en puissance du mobile ? Les applications natives plus plébiscitées sur mobileLe web serait-il en danger face à la montée en puissance du mobile ? Oui, pour le chercheur et écrivain Danny Crichton, pour qui le web se meurt lentement et surement à l'aube de son 25e anniversaire.Tout d'abord, il est important de définir le web, qui représente « une collection de protocoles (HTP) et de langages (HTML) qui permet à l'utilisateur de produire et de consommer...

    Read the article

  • Why do I have an error when adding states in slick?

    - by SystemNetworks
    When I was going to create another state I had an error. This is my code: public static final int play2 = 3; and public Game(String gamename){ this.addState(new mission(play2)); } and public void initStatesList(GameContainer gc) throws SlickException{ this.getState(play2).init(gc, this); } I have an error in the addState. above the above code. I don't know where is the problem. But if you want the whole code it is here: package javagame; import org.newdawn.slick.*; import org.newdawn.slick.state.*; public class Game extends StateBasedGame{ public static final String gamename = "NET FRONT"; public static final int menu = 0; public static final int play = 1; public static final int train = 2; public static final int play2 = 3; public Game(String gamename){ super(gamename); this.addState(new Menu(menu)); this.addState(new Play(play)); this.addState(new train(train)); this.addState(new mission(play2)); } public void initStatesList(GameContainer gc) throws SlickException{ this.getState(menu).init(gc, this); this.getState(play).init(gc, this); this.getState(train).init(gc, this); this.enterState(menu); this.getState(play2).init(gc, this); } public static void main(String[] args) { try{ AppGameContainer app =new AppGameContainer(new Game(gamename)); app.setDisplayMode(1500, 1000, false); app.start(); }catch(SlickException e){ e.printStackTrace(); } } } //SYSTEM NETWORKS(C) 2012 NET FRONT

    Read the article

  • Entity Relationship diagram - Composition

    - by GigaPr
    Hi, I am implementing a small database(university Project) and i am facing the following problem. I created a class diagram where i have a class Train {Id, Name, Details} And a class RollingStock which is than generalized in Locomotive and FreightWagon. A train is Composed by multiple RollingStock at a certain time(on different days the rolling stock will compose a different train). I represented the relationship train - rolling stock as a diamond filled (UML) but still I have a many to many relationship between the two tables. so i guess i have to create an additional table to solve the many to many relationship train_RollingStock. but how do i represent the Composition? Can i still use the filled diamond? If yes on which side? Thanks

    Read the article

  • making apache and django add a trailing slash

    - by user302099
    Hello. My /train directory is aliased to a script in httpd.conf by: WSGIScriptAlias /train /some-path/../django.wsgi And it works well, except for one problem. If a user goes to /train (with no trailing slash) it will not redirect him to /train/, but will just give him the right page. This is a problem because this way the relative links on this page lead to the wrong place when no trailing slash was used to access it. How can this be worked out? Thanks.

    Read the article

  • Ruby On Rails: Ask for Confirmation When Table Entry Associated With Another Is Destroyed

    - by Train Main
    Hi all, I would like some assistance with the following problem: I have a table of groups that is self-associated with itself, so each group is (optionally) linked to another in a hierarchical fashion. I want to write some code that will somehow check before the destruction of a group entry, if it has any children, and ask the user for confirmation, or whether they wish to delete the child groups as well. I've looked at callbacks, but I don't know how to get the confirmation request to the end user in the view, and then get the response back to the model's callback. Thanks

    Read the article

  • C# Neural Networks with Encog

    - by JoshReuben
    Neural Networks ·       I recently read a book Introduction to Neural Networks for C# , by Jeff Heaton. http://www.amazon.com/Introduction-Neural-Networks-C-2nd/dp/1604390093/ref=sr_1_2?ie=UTF8&s=books&qid=1296821004&sr=8-2-spell. Not the 1st ANN book I've perused, but a nice revision.   ·       Artificial Neural Networks (ANNs) are a mechanism of machine learning – see http://en.wikipedia.org/wiki/Artificial_neural_network , http://en.wikipedia.org/wiki/Category:Machine_learning ·       Problems Not Suited to a Neural Network Solution- Programs that are easily written out as flowcharts consisting of well-defined steps, program logic that is unlikely to change, problems in which you must know exactly how the solution was derived. ·       Problems Suited to a Neural Network – pattern recognition, classification, series prediction, and data mining. Pattern recognition - network attempts to determine if the input data matches a pattern that it has been trained to recognize. Classification - take input samples and classify them into fuzzy groups. ·       As far as machine learning approaches go, I thing SVMs are superior (see http://en.wikipedia.org/wiki/Support_vector_machine ) - a neural network has certain disadvantages in comparison: an ANN can be overtrained, different training sets can produce non-deterministic weights and it is not possible to discern the underlying decision function of an ANN from its weight matrix – they are black box. ·       In this post, I'm not going to go into internals (believe me I know them). An autoassociative network (e.g. a Hopfield network) will echo back a pattern if it is recognized. ·       Under the hood, there is very little maths. In a nutshell - Some simple matrix operations occur during training: the input array is processed (normalized into bipolar values of 1, -1) - transposed from input column vector into a row vector, these are subject to matrix multiplication and then subtraction of the identity matrix to get a contribution matrix. The dot product is taken against the weight matrix to yield a boolean match result. For backpropogation training, a derivative function is required. In learning, hill climbing mechanisms such as Genetic Algorithms and Simulated Annealing are used to escape local minima. For unsupervised training, such as found in Self Organizing Maps used for OCR, Hebbs rule is applied. ·       The purpose of this post is not to mire you in technical and conceptual details, but to show you how to leverage neural networks via an abstraction API - Encog   Encog ·       Encog is a neural network API ·       Links to Encog: http://www.encog.org , http://www.heatonresearch.com/encog, http://www.heatonresearch.com/forum ·       Encog requires .Net 3.5 or higher – there is also a Silverlight version. Third-Party Libraries – log4net and nunit. ·       Encog supports feedforward, recurrent, self-organizing maps, radial basis function and Hopfield neural networks. ·       Encog neural networks, and related data, can be stored in .EG XML files. ·       Encog Workbench allows you to edit, train and visualize neural networks. The Encog Workbench can generate code. Synapses and layers ·       the primary building blocks - Almost every neural network will have, at a minimum, an input and output layer. In some cases, the same layer will function as both input and output layer. ·       To adapt a problem to a neural network, you must determine how to feed the problem into the input layer of a neural network, and receive the solution through the output layer of a neural network. ·       The Input Layer - For each input neuron, one double value is stored. An array is passed as input to a layer. Encog uses the interface INeuralData to hold these arrays. The class BasicNeuralData implements the INeuralData interface. Once the neural network processes the input, an INeuralData based class will be returned from the neural network's output layer. ·       convert a double array into an INeuralData object : INeuralData data = new BasicNeuralData(= new double[10]); ·       the Output Layer- The neural network outputs an array of doubles, wraped in a class based on the INeuralData interface. ·        The real power of a neural network comes from its pattern recognition capabilities. The neural network should be able to produce the desired output even if the input has been slightly distorted. ·       Hidden Layers– optional. between the input and output layers. very much a “black box”. If the structure of the hidden layer is too simple it may not learn the problem. If the structure is too complex, it will learn the problem but will be very slow to train and execute. Some neural networks have no hidden layers. The input layer may be directly connected to the output layer. Further, some neural networks have only a single layer. A single layer neural network has the single layer self-connected. ·       connections, called synapses, contain individual weight matrixes. These values are changed as the neural network learns. Constructing a Neural Network ·       the XOR operator is a frequent “first example” -the “Hello World” application for neural networks. ·       The XOR Operator- only returns true when both inputs differ. 0 XOR 0 = 0 1 XOR 0 = 1 0 XOR 1 = 1 1 XOR 1 = 0 ·       Structuring a Neural Network for XOR  - two inputs to the XOR operator and one output. ·       input: 0.0,0.0 1.0,0.0 0.0,1.0 1.0,1.0 ·       Expected output: 0.0 1.0 1.0 0.0 ·       A Perceptron - a simple feedforward neural network to learn the XOR operator. ·       Because the XOR operator has two inputs and one output, the neural network will follow suit. Additionally, the neural network will have a single hidden layer, with two neurons to help process the data. The choice for 2 neurons in the hidden layer is arbitrary, and often comes down to trial and error. ·       Neuron Diagram for the XOR Network ·       ·       The Encog workbench displays neural networks on a layer-by-layer basis. ·       Encog Layer Diagram for the XOR Network:   ·       Create a BasicNetwork - Three layers are added to this network. the FinalizeStructure method must be called to inform the network that no more layers are to be added. The call to Reset randomizes the weights in the connections between these layers. var network = new BasicNetwork(); network.AddLayer(new BasicLayer(2)); network.AddLayer(new BasicLayer(2)); network.AddLayer(new BasicLayer(1)); network.Structure.FinalizeStructure(); network.Reset(); ·       Neural networks frequently start with a random weight matrix. This provides a starting point for the training methods. These random values will be tested and refined into an acceptable solution. However, sometimes the initial random values are too far off. Sometimes it may be necessary to reset the weights again, if training is ineffective. These weights make up the long-term memory of the neural network. Additionally, some layers have threshold values that also contribute to the long-term memory of the neural network. Some neural networks also contain context layers, which give the neural network a short-term memory as well. The neural network learns by modifying these weight and threshold values. ·       Now that the neural network has been created, it must be trained. Training a Neural Network ·       construct a INeuralDataSet object - contains the input array and the expected output array (of corresponding range). Even though there is only one output value, we must still use a two-dimensional array to represent the output. public static double[][] XOR_INPUT ={ new double[2] { 0.0, 0.0 }, new double[2] { 1.0, 0.0 }, new double[2] { 0.0, 1.0 }, new double[2] { 1.0, 1.0 } };   public static double[][] XOR_IDEAL = { new double[1] { 0.0 }, new double[1] { 1.0 }, new double[1] { 1.0 }, new double[1] { 0.0 } };   INeuralDataSet trainingSet = new BasicNeuralDataSet(XOR_INPUT, XOR_IDEAL); ·       Training is the process where the neural network's weights are adjusted to better produce the expected output. Training will continue for many iterations, until the error rate of the network is below an acceptable level. Encog supports many different types of training. Resilient Propagation (RPROP) - general-purpose training algorithm. All training classes implement the ITrain interface. The RPROP algorithm is implemented by the ResilientPropagation class. Training the neural network involves calling the Iteration method on the ITrain class until the error is below a specific value. The code loops through as many iterations, or epochs, as it takes to get the error rate for the neural network to be below 1%. Once the neural network has been trained, it is ready for use. ITrain train = new ResilientPropagation(network, trainingSet);   for (int epoch=0; epoch < 10000; epoch++) { train.Iteration(); Debug.Print("Epoch #" + epoch + " Error:" + train.Error); if (train.Error > 0.01) break; } Executing a Neural Network ·       Call the Compute method on the BasicNetwork class. Console.WriteLine("Neural Network Results:"); foreach (INeuralDataPair pair in trainingSet) { INeuralData output = network.Compute(pair.Input); Console.WriteLine(pair.Input[0] + "," + pair.Input[1] + ", actual=" + output[0] + ",ideal=" + pair.Ideal[0]); } ·       The Compute method accepts an INeuralData class and also returns a INeuralData object. Neural Network Results: 0.0,0.0, actual=0.002782538818034049,ideal=0.0 1.0,0.0, actual=0.9903741937121177,ideal=1.0 0.0,1.0, actual=0.9836807956566187,ideal=1.0 1.0,1.0, actual=0.0011646072586172778,ideal=0.0 ·       the network has not been trained to give the exact results. This is normal. Because the network was trained to 1% error, each of the results will also be within generally 1% of the expected value.

    Read the article

  • Floating point undesireable in highly critical code?

    - by Kirt Undercoffer
    Question 11 in the Software Quality section of "IEEE Computer Society Real-World Software Engineering Problems", Naveda, Seidman, lists fp computation as undesirable because "the accuracy of the computations cannot be guaranteed". This is in the context of computing acceleration for an emergency braking system for a high speed train. This thinking seems to be invoking possible errors in small differences between measurements of a moving object but small differences at slow speeds aren't a problem (or shouldn't be), small differences between two measurements at high speed are irrelevant - can there be a problem with small roundoff errors during deceleration for an emergency braking system? This problem has been observed with airplane braking systems resulting in hydroplaning but could this actually happen in the context of a high speed train? The concern about fp errors seems to not be well-founded in this context. Any insight? The fp is used for acceleration so perhaps the concern is inching over a speed limit? But fp should be just fine if they use a double in whatever implementation language. The actual problem in the text states: During the inspection of the code for the emergency braking system of a new high speed train (a highly critical, real-time application), the review team identifies several characteristics of the code. Which of these characteristics are generally viewed as undesirable? The code contains three recursive functions (well that one is obvious). The computation of acceleration uses floating point arithmetic. All other computations use integer arithmetic. The code contains one linked list that uses dynamic memory allocation (second obvious problem). All inputs are checked to determine that they are within expected bounds before they are used.

    Read the article

  • Floating point undesirable in highly critical code?

    - by Kirt Undercoffer
    Question 11 in the Software Quality section of "IEEE Computer Society Real-World Software Engineering Problems", Naveda, Seidman, lists fp computation as undesirable because "the accuracy of the computations cannot be guaranteed". This is in the context of computing acceleration for an emergency braking system for a high speed train. This thinking seems to be invoking possible errors in small differences between measurements of a moving object but small differences at slow speeds aren't a problem (or shouldn't be), small differences between two measurements at high speed are irrelevant - can there be a problem with small roundoff errors during deceleration for an emergency braking system? This problem has been observed with airplane braking systems resulting in hydroplaning but could this actually happen in the context of a high speed train? The concern about fp errors seems to not be well-founded in this context. Any insight? The fp is used for acceleration so perhaps the concern is inching over a speed limit? But fp should be just fine if they use a double in whatever implementation language. The actual problem in the text states: During the inspection of the code for the emergency braking system of a new high speed train (a highly critical, real-time application), the review team identifies several characteristics of the code. Which of these characteristics are generally viewed as undesirable? The code contains three recursive functions (well that one is obvious). The computation of acceleration uses floating point arithmetic. All other computations use integer arithmetic. The code contains one linked list that uses dynamic memory allocation (second obvious problem). All inputs are checked to determine that they are within expected bounds before they are used.

    Read the article

  • Save PyML.classifiers.multi.OneAgainstRest(SVM()) object?

    - by Michael Aaron Safyan
    I'm using PYML to construct a multiclass linear support vector machine (SVM). After training the SVM, I would like to be able to save the classifier, so that on subsequent runs I can use the classifier right away without retraining. Unfortunately, the .save() function is not implemented for that classifier, and attempting to pickle it (both with standard pickle and cPickle) yield the following error message: pickle.PicklingError: Can't pickle : it's not found as __builtin__.PySwigObject Does anyone know of a way around this or of an alternative library without this problem? Thanks. Edit/Update I am now training and attempting to save the classifier with the following code: mc = multi.OneAgainstRest(SVM()); mc.train(dataset_pyml,saveSpace=False); for i, classifier in enumerate(mc.classifiers): filename=os.path.join(prefix,labels[i]+".svm"); classifier.save(filename); Notice that I am now saving with the PyML save mechanism rather than with pickling, and that I have passed "saveSpace=False" to the training function. However, I am still gettting an error: ValueError: in order to save a dataset you need to train as: s.train(data, saveSpace = False) However, I am passing saveSpace=False... so, how do I save the classifier(s)? P.S. The project I am using this in is pyimgattr, in case you would like a complete testable example... the program is run with "./pyimgattr.py train"... that will get you this error. Also, a note on version information: [michaelsafyan@codemage /Volumes/Storage/classes/cse559/pyimgattr]$ python Python 2.6.1 (r261:67515, Feb 11 2010, 00:51:29) [GCC 4.2.1 (Apple Inc. build 5646)] on darwin Type "help", "copyright", "credits" or "license" for more information. import PyML print PyML.__version__ 0.7.0

    Read the article

  • Implementing simple business model in Haskell

    - by elmes
    Supose we have a very simple model: Station has at least one Train Train has at least two Stations The model has to allow to check what stations any given train visits and to check what trains visit a particular station. How to model it in Haskell? I am a Haskell newbie, so please correct me: once an object is created, you cannot modify it - you can only make a new object based on that one (~immutability). Am I right? If so, I'll have to create a lot of temporary variables with semi-initialized objects (during deserialization or even in unit tests). Basically what I need is an example of modeling domain classes in Haskell - after reading "Learn you a haskell.." I still have no idea how to use this language.

    Read the article

  • R problems using rpart with 4000 records and 13 attributes

    - by josh
    I have attempted to email the author of this package without success, just wondering if anybody else has experienced this. I am having an using rpart on 4000 rows of data with 13 attributes. I can run the same test on 300 rows of the same data with no issue. When I run on 4000 rows, Rgui.exe runs consistently at 50% cpu and the UI hangs.... it will stay like this for at least 4-5hours if I let it run, and never exit or become responsive. here is the code I am using both on the 300 and 4000 size subset : train<-read.csv("input.csv",header=T) y<-train[,18] x<-train[,3:17] library(rpart) fit<-rpart(y~.,x) Is this a known limitation of rpart, am I doing something wrong? potential workarounds? any assistance appreciated

    Read the article

  • Grounded in Dublin

    - by Mike Dietrich
    Friday's hands-on workshop in the Oracle office in Dublin was quite good fun for everybody - except for Mick who has just recognized that his Ryanair flight back to Cork has been canceled (So I hope you've returned home well!) and me as my flights back to Munich via London City had been canceled as well. It's always good to have somebody in the workshop from Air Lingus so I've got hourly information what's going in in the Irish airspace (and now I know that the system dealing with such situations is an well prepared Oracle database which runs just like a switch watch - Thanks again for all your support!!! Was great to talk to you!!!). But to be honest, there are worse places to be grounded for a few days than Dublin. At least it gave me the chance to do something which I never had time enough before when visiting Oracle Ireland: a bit of sightseeing. When I've realized that nothing seems to move over the weekend I started organizing my travel back yesterday. It was no fun at all because there's no single system to book such a travel. Figuring out all possibilities and options going back to Munich was the first challange. Irish Ferries webpage was moaning with all the unexpected load (currently it's fully down). Hotel booking websites showed vacancies in Holyhead but didn't let me book. And calling them just reveiled that there are no rooms left. Haven't stayed overnight in a train station for quite a while ;-) The website of VirginTrains puzzled me with offering a seat at an enormous price for a train ride from Holyhead to London Euston (Thanks, Sir Richard Branson!) just to tell me after I booked a ticket that there are no seats left (but I traveled German railsways a few weeks ago from Düsseldorf to Frankfurt sitting on the floor as well). Eurostar's website let me choose tickets through the tunnel to tell me in the final step that the ticket cannot be confirmed as there are no seats left - but the next check again showed bookable seats - must be a database from some other vendor which has no proper row level locking ... hm ...?! Finally the TGV page for the speed train to Stuttgart and then the ICE to Munich was not allowing searches for quite a while - but ultimately ... after 4.5 hours of searching, waiting, sending credit card information again and again ... So if you have a few spare fingers please keep them crossed :-) And good luck to all my colleagues traveling back from the Exadata training in Berlin. As Mike Appleyard, my colleague from the UK presales team wrote: "Dublin and Berlin aren't too bad a place to get stuck... ;-)"

    Read the article

  • Back home :-)

    - by Mike Dietrich
    Wrote this entry last night in the ICE from Stuttgart to Munich but the conncetion broke: 28.5 hour journey - and close by now. Actually I would have been even closer if our TGV wouldn't have had break problems as soon as we had entered German territory. And you don't want a train which goes up to a speed of 200 mph having issues with its breaks, right? So we missed the connection in Stuttgart but I've catched the last train this night towards Munich. Distance approx 1900 km all together. Usually it takes 2.5 hours with a direct flight with Air Lingus from Munich or a bit more when you'll go through Zurich or Frankfurt. But at least you meet more people and see a bit more from the landscapes passing by :-) Except for the break problem everything worked out well so far (I'm no there finally!). I had 4 hours to change in Paris from Gare de Nord to Gare de l'Est and one thing I really have to point out: the people working for SNCF, the French National Railways, were so organized and helpful, purely amazing. I asked the man at the counter where I had to pick up my prepaid tickets for directions to Gare de l'Est - and after we had a chat about Marlene Dietrich he just grabbed his iPhone, started Google Earth and showed me the way to walk. I pretty sure it's a stupid stereotype that people in Paris or France are so unfriendly to foreigners if they don't speak French. In my past 3 stays or travels to Paris in the past 2 years I had only great experiences. And another thing I really enjoy when being in France: the food!!! The sandwich I had at the train station was packed with yummy goat cheese. And there's always Paul. You might ask yourself: Who the heck is Paul? That's Paul - or actually their website. And at Paul's they serve usually excellent fruit tartes - and this time a nice Gateau Au Chocolate. And very good Cafe Cremé as well :-) That's actually the positive part traveling this way: the food you'll get is much better than the airline food - if your airline still serves something called food ...

    Read the article

  • Craftsmanship Tour: Day 2 Obtiva

    - by Liam McLennan
    I like Chicago. It is a great city for travellers. From the moment I got off the plane at O’Hare everything was easy. I took the train to ‘the Loop’ and walked around the corner to my hotel, Hotel Blake on Dearborn St. Sadly, the elevated train lines in downtown Chicago remind me of ‘Shall We Dance’. Hotel Blake is excellent (except for the breakfast) and the concierge directed me to a pizza place called Lou Malnati's for Chicago style deep-dish pizza. Lou Malnati’s would be a great place to go with a group of friends. I felt strange dining there by myself, but the food and service were excellent. As usual in the United States the portion was so large that I could not finish it, but oh how I tried. Dave Hoover, who invited me to Obtiva for the day, had asked me to arrive at 9:45am. I was up early and had some time to kill so I stopped at the Willis Tower, since it was on my way to the office. Willis Tower is 1,451 feet (442 m) tall and has an observation deck at the top. Around the observation deck are a set of acrylic boxes, protruding from the side of the building. Brave soles can walk out on the perspex and look between their feet all the way down to the street. It is unnerving. Obtiva is a progressive, craftsmanship-focused software development company in downtown Chicago. Dave even wrote a book, Apprenticeship Patterns, that provides a catalogue of patterns to assist aspiring software craftsmen to achieve their goals. I spent the morning working in Obtiva’s software studio, an open xp-style office that houses Obtiva’s in-house development team. For lunch Dave Hoover, Corey Haines, Cory Foy and I went to a local Greek restaurant (not Dancing Zorbas). Dave, Corey and Cory are three smart and motivated guys and I found their ideas enlightening. It was especially great to chat with Corey Haines since he was the inspiration for my craftsmanship tour in the first place. After lunch I recorded a brief interview with Dave. Unfortunately, the battery in my camera went flat so I missed recording some interesting stuff. Interview with Dave Hoover In the evening Obtiva hosted an rspec hackfest with David Chelimsky and others. This was an excellent opportunity to be around some of the very best ruby programmers. At 10pm I went back to my hotel to get some rest before my train north the next morning.

    Read the article

facebook twitter digg google delicious technorati stumbleupon myspace wordpress linkedin gmail igoogle windows live tumblr viadeo yahoo buzz yahoo mail yahoo bookmarks favorites email print

< Previous Page | 1 2 3 4 5 6 7 8 9 10 11 12  | Next Page >