December 2003 - cypherpunks-legacy

Re: faster code for neural network
by Matthew Bernardini 17 Dec '03

17 Dec '03

> >> Recently I received the code for a neural network. It is written in > >> standard ANSI-C. However, running it on the Mac gives me a lot of coffee > >> breaks ( it takes 10 min on a IIci, 3 hrs on an LC ). I was wondering if > >> anybody out there has experience with this and knows of ways to make the > >> code faster ( replacing ANSI routines with Mac-based routines. > >> > >> If you are the woman or man with this experience mail me! I really need the > >> help ( it's all for my master's thesis ). > > > >You might want to get a match coprocessor for the LC. The ci has one, > >and that's probably your problem. (I'm not sure if the LC can take one, > >though!) If my memory still works after the two pints I just had, the LC > >has a 68020 without any FPU. You might want to upgrade to an LC III and > >find out if it has an FPU. I believe neural nets use floats, no? If so > >you will most certainly need and FPU if you want decent speed. > > > > The LC does lack a FPU, but you can get one on a card. I'm not sure if > this would help though, because it would depend if your compilier would > generate FPU code. I know you can with Think. You also might need to take > into account which compiler you are using for this kind of program, there > seems to be a vairation in efficency between Mac Compilers. > > As for using toolbox calls, using Apple's SANE library might help. I > haven't had much experience with it, but it would probably take advatage of > the FPU. > > E-mail me if I can help anymore. (I'll run the program on my Q700 if you > just need a quick speed test.) > ___________________________________________________________________________ > "I want to know God's thoughts...the rest are details." > -- Albert Einstein > _________________________________________ > Ron Davis rondavis(a)datawatch.com > Datawatch, Research Triangle Park, NC (919)549-0711 > > That, or use an AV. If you are adventurous you could use Aplle's Real Time Architecture to offload processing to the 66MHZ DSP processor in the AV series. I know of some examples of ARTA. Matthew Bernardini ----------------------------------------------------------------------------- | Rutgers University Computing Services Matthew Bernardini | Hill Micro/Graphics Center 7804 McCormick | Site-Manager (908) 878-0946 | 017 Hill Center | (908) 932-3129 (908) 932-4921 -----------------------------------------------------------------------------

1 0

Re: AI code speed up !!!!
by Scott Bence 17 Dec '03

17 Dec '03

I don't know what (if) I read this in any books or mags. I would recommend MacTutor(name may have changed) I had read it all through college and worshiped it. You can find it at *the Book Stop* if you know where on is. You can benchmark you routines fairly easy. Before you call your function get the ticks off the clock then again afterwards, then get the difference. ie ... start=GetTicks(); /* I think this was call, check the toolbox */ foobar(); total_time=GetTicks()-start; ... One more thing, I left out another optimization kludge. Try to reduce the number of calls outside your function by moving the code it is calling to inside the function. This reduces the associated overhead of jumping (jsr $xxxx). Also, if you have to go through the "toolbox" check into the function GetTrapAddress(). Do this for Moveto(x,y) and then you can bypass the entry point of the toolbox and go directly to the routine in ROM. This should shave off 5%-15% on the execution of toolbox calls. DO NOT HARD CODE THE ADDRESS IN YOUR PROGRAM! YOUR PROGRAM WILL DIE A HORRIBLE DEATH ON ANY OTHER MODEL! Use GetTrapAddress(). Hope all this helps. Let me know how things work for you. > Bence, > > I saw your list of recommendations for code speed-up on the mac. I'm > currently learning how to program the mac, make tool-box calls, etc. and > I found the list interesting. Is there book of information like that, or > is this just stuff picked up in the trade? Let me know when you get a chance. > > -john. > jdblair(a)nextsrv.cas.muohio.edu > #include <std_disclaimer.h> ---------------------------------------------------------------------- ------ Scott Bence * Texas State Technical College @ Waco, Texas ----- --- Computer Networking & Systems Administration * sbence(a)tstc.edu --- ----------------------------------------------------------------------

1 0

Re: alternative to Fair Cryptosystems
by fb＠cyberg.win.net 17 Dec '03

17 Dec '03

>You can have your surveillance agency (or agencies which need to cooperate) >publish their own RSA keys (big ones, presumably), and all the good little >boys and girls who want to prove how obedient and conformist they are can >include those keys as recipients when they encrypt messages. If there are >to be multiple agencies which have to cooperate, the PGP or RIPEM software >would have to change to split the message key by XOR with ranno pieces, >but in the meantime, you could just include the FBI in your list of recipients >and save everybody the hassle of having to get pieces to put together. > >Simple -- direct -- speaks right to the gov't desire. What could be wrong >with this? And anyone who didn't include a surveillance agency in the list of message recipients would of course be demonstrating that they were up to no good and should be investigated. How diabolically clever. Next let's "voluntarily" switch to postcards for all snail mail. :) --------------------------------------------------------------- Francis Barrett, F.R.C. | Thou canst not travel on the path | The Cybernetics Guild | before thou hast become the Path | fb(a)cyberg.win.net | itself. | ---------------------------------------------------------------

1 0

Re: Internet Banking
by Matthew Bernardini 17 Dec '03

17 Dec '03

> (First of all, I just wanted to say hello, especially to those of you I > met for the first time at the Cypherpunks BOF last night at Usenix.) > > I caught up on the digital banking thread and have a few thoughts > about future banks, financial privacy, and nonregulation issues, > especially how they relate to the way things are done now. > > What is important to me in an ideal bank, in roughly descending order: > > 1. Security. The bank will not disappear, steal my money, allow others > to steal my money, or unwarrantedly cut off my access to it. > > 2. Convenience. I can easily and reliably spend the money I have on deposit, > deposit funds from elsewhere, and communicate with the bank regarding > my accounts. > > 3. Privacy. The bank will not (or, possibly, cannot) reveal details > of my financial situation or transactions to others including the > government. > > 4. Credit. The bank will loan me money for appropriate purposes if > needed (and my credit rating is acceptable). > > The financial institutions I currently deal with do a pretty good job > of all of these except (3), which is not their fault but is the > government's. (Except for crud like banks that use trivial keys like > your SSN for access to banking by phone.) > > No. 1, security, is a problem with anonymized, offshore, network > banking. Today we rely on a combination of reputation and regulation > to provide bank security -- the banks we deal with stress size, > longevity, permanence, etc., in their marketing campaigns, and there > are mandatory reserve requirements and mandatory deposit insurance. > Reputation should translate pretty well in our idealized banking world > -- what is better than the electronic word of mouth of the Internet? > But in a nonregulated environment, there will have to be private > deposit insurance which could easily have some bootstrap problems in > building the initial market. > > Convenience, #2, should be a vast improvement. Freed from the > necessity and cost of maintaining a network of impressive physical > edifices of Federal-style architecture, and coupled with more-or-less > ubiquitous networking and computing, banks can concentrate on giving > ultra-fast, efficient transaction services via authenticated e-mail > and customer services via a Web-like server. > > I envision transactions ending up in two big buckets: card services > and "cheque" services. Cards are for when you are wandering around, > and e-mail "cheques" are for paying regular bills. The card system, > insecure at it is, is fast becoming universal; I stopped carrying a > checkbook around years ago and use credit cards for all possible > transactions: you get a comprehensive statement at the end of the month with > the names of all your vendors, and you also get a nice premium for > using their transaction services (mine is airline frequent-flyer miles). > In any future banking system one must assume that card-based > transaction service will be the main, if not only, means of > casual transaction, and it will be up to us to to build in the > ncessary privacy and authenticating schemes to make this a trustable > system. I find it difficult to imagine large-scale displacement of > institutions like VISA, MasterCard, and Amex, simply because they do > what they do (provide instant POS credit authorization, guarantee > merchants quick payment, etc.) very well. > > While e-mail "cheques" are attractive because the mail infrastructure > is almost entirely there already, I wonder if they will ever become more > than a small percentage of total transactions, possibly limited to > pre-authorized direct drafts for such things as utility bills, and > maybe mail orders and transactions between individuals. > > #3, privacy, is a very difficult issue because of the regulatory role > of the government. Because of the degree to which strong financial > privacy threatens government power (especially taxing power; see > previous messages on this) I can easily imagine that (1) banks doing > business in or "touching" the US and most politically similar > sovereignties will not be able to prevent themselves from disclosing > identity and transaction information about their customers, and (2) > people will probably be prohibited from dealing with these banks if > they are in fact beyond the reach of legal process. > > This leads to the key question: should one trust (i.e., disclose > one's identity to) or not trust one's bank? It would certainly be nice > to be able to trust your bank, as it makes things much easier for all > parties. They would be able to freely grant you credit (#4 above), > since you could verifiably prove your assets, real property, > employment, etc. But if you trust your bank, then they may be forced > to disclose your identity to the government under legal process. Thus > it is probably best to postulate a banking system that does not > require trust. > > This complicates #4, credit. I can envision a system of vouched-for > indirection (not unlike signing PGP-keys) that would allow you (the > borrower) to disclose assets/earning capacity to a trusted third party > that would certify to a lender that you (known to the lender only as a > numbered account) are credit-worthy. If it can be made possible > to break the traceability link between the credit-vouching agency and > the lender, privacy may be possible. One problem may be that > credit-vouching agencies cannot easily be "offshore", since they may > need to examine your real estate (or whatever) though this could be > done, perhaps, by appraisers or other local agents. > > Comments? > > -- > Michael C. Berch > mcb(a)net.bio.net / mcb(a)postmodern.com / mcb(a)remarque.berkeley.edu > > I am new to this list, so excuse me if this topic has already been discussed, but I think you need to take a 200 level course in economics called Money and Banking. I think the idea is so obsessed with tax-evasion and privacy protection that you have ignored all the economic consequences of the ideas you are proposing. 1) Who will insure your money ? Can you trust anyone but the US gov't to back your funds ? Even in the S&L scandal the gov't refunded money to people that weren't insured by the FDIC. Do you think they would come to the cypher-punk rescue if your money up and flew to Brazil ? 2) A doctoral thesis could be written about this one, but what about the Federal Reserve ? You would wreak havoc on interest rates, inflation, international balance of payments, and international trade. How would this electronic bank adjust for inflation or an expanding/shrinking electronic money supply ? Take a look at some historical texts that describe the problems that the Early American Revolutionaries had in breaking from the British Currency. It took several failed efforts, and the currency of the United States has been constatnly evolving ever since. 3) Interest Rates and Inflation ... 4) Interest Rates and Inflation .... 5) You guessed it, Interest Rates and Inflation. What about Capital Markets ? What about foreign labor unit exchanges? Is mexican labor worth as much as US labor? I think there are more economic issues than encryption issues to be considered in this case. Matt ----------------------------------------------------------------------------- | Rutgers University Computing Services Matthew Bernardini | Hill Micro/Graphics Center 7804 McCormick | Site-Manager (908) 878-0946 | 017 Hill Center | (908) 932-3129 (908) 932-4921 -----------------------------------------------------------------------------

2 1

Off-Shore Banking
by DECLAN＠delphi.com 17 Dec '03

17 Dec '03

Complete guides to off-shore banking, and more, are available from LOW PROFILE newsletter POBox 84910 Phoenizx, Arizona 85701 Tel 1-800-528-0559 Fax 602-943-2363 I subscribe to this newsletter and find the information well-researched and worthwhile. If a banking institution has a bad reputation, they publish the information. Lots of other information on asset ptrotection at home and abroad. declan(a)delphi.com

1 0

RE: RSA is weak link
by Karl Lui Barrus 17 Dec '03

17 Dec '03

-----BEGIN PGP SIGNED MESSAGE----- > ...the IDEA encryption algorithm used in PGP is actually > MUCH stronger than RSA given the same key length. Even with a 1024 bit > RSA key, it is believed that IDEA encryption is still stronger, and, > since a chain is no stronger than it's weakest link, it is believed that > RSA is actually the weakest part of the RSA - IDEA approach. >Confirmation? True; it is more expensive to do a brute force search over the IDEA keyspace than it is to factor a 1024 bit number using the best known factoring algorithm :) I did some calculations once and I think the crossover point is between 1500 and 1600 bits, at which point factoring becomes more difficult than searching. -----BEGIN PGP SIGNATURE----- Version: 2.3a iQCVAgUBLUA1OoOA7OpLWtYzAQGjwwQAwhDqTc6TKZqEjfO1xbC1llasfVfOL3wM 72ofDeWtw7vtdxFb98iLShomdwr12G6I+7qbHwzKyLM5+r/T/DzNRAW2e+MD894Y DNAZazRz2YLuiGpr0LBSC1S6EyFbtcuwoPh1U5LORyUrEo4YCJsYYoFfcnN+Vb8B ddv28VkktBs= =irY2 -----END PGP SIGNATURE-----

1 0

Knowledgeable people in the Boston area
by Alan (Miburi-san) Wexelblat 17 Dec '03

17 Dec '03

I know that some members of this list are in the Boston area; I'm also wondering if any of the "names" in crypto live up here. I'm trying to arrange a talk (elementary level) for a group of human-interface designers I chair and am hunting up speaker ideas. I will be out of the office tomorrow, but will respond to emails Monday. Thanks! --Alan Wexelblat, Reality Hacker, Author, and Cyberspace Bard Media Lab - Advanced Human Interface Group wex(a)media.mit.edu Voice: 617-258-9168 Page: 617-945-1842 an53607(a)anon.penet.fi All the world's a stage and most of us are desperately unrehearsed.

1 0

unsubscribe plea
by Wendy Whoppers 17 Dec '03

17 Dec '03

sorry, I KNOW that unsubscribe messages should be routed to -request but I have tried that many many times in all possible combinations to no avail. Anyone know the secret password to unsubscribe? Muchas, and sorry for the wasted space. --Joel <growjc(a)whitman.edu>

1 0

Re: APPLIED CRYPTOGRAPHY - Index
by norm＠netcom.com 17 Dec '03

17 Dec '03

At 11:12 1/19/94 -0600, Bruce Schneier wrote: ...etc. ... Thanks for the index. You probably sold another book here.

1 0

DeTwEiLeR IdIoCy (was Re: SQUISH IDIOCY)
by lefty＠apple.com 17 Dec '03

17 Dec '03

"Leisure Suit" Larry Detweiler writes: > >Do YOU really think that *anything* can be posted without ANYONE being >accountable? YOU cypherpunks, YOU are nothing but drenched, permeated, >rotton, noxious IDIOTS. Even J. Helsingius, operator of the most >reliable anonymous site ever, has apparently censored S.Boxx in >response to YOUR pressure. Doesn't it bother you to realize that your >most treasured beliefs, the most sacred elements of your religion, are >nothing but LIES? YOU have clung to the illusion that no one is >accountable for anything, when in fact YOU VIOLATE and DEBASE YOUR OWN >SACRED RELIGION. Heh. Detweiler as McElwaine. Pull the other one, Larry. -- Lefty (lefty(a)apple.com) C:.M:.C:., D:.O:.D:.

1 0