Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Comment Re:But... (Score 2) 139

It is indeed true that one neural network was trained using a collection of games, but the version of AlphaGo that played against Lee Sedol last year, was using two neural networks, and the second one (for evaluating the positional strength of a board configuration) was trained by letting AlphaGo play against itself. It is not known how he current version of AlphaGo works, whether any additional neural networks were added, but if it has become stronger, it has done so by playing against itself. It should be noted that this latest version of AlphaGo was playing some surprising moves that made some people believe, it could not have been AlphaGo, or that it was a version that was trained without using a collection of games. It should be noted that DeepMind announced it wanted to experiment with training a neural network with zero additional knowledge for the game of go, just like they did with the neural networks playing old video games.

Comment Re:Typical (Score 5, Insightful) 139

You seemed have to missed the fact that many of the top professional players were lining up to play against this bot. They view it as free training lession, not to beat an AI bot, but to beat their human counter parts. Since Lee Sedol played against AlphaGo, he has gained in strength, so much even that a certain point, using a certain method, AlphaGo was the strongest player, not because it had played more games, but because Lee Sedol had won so many games. Ke Jie, to be considered the strongest player at the moment, has made remarks that humans have only touched at the truth behind go, after he played against Master(P). Most go players have a very high regard for the game, as they sense that it is much deeper than human mind can consider. For this reason, I guess, many professional go players find this a very exicting time, because it will enhance their understanding of the game. In this view it is very unlikely that a professional player will use a trick to force a tie.

Comment Use unparser (Score 1) 108

I would let LISP print out a s-expression representing the abstract syntax tree (usually called abstract parse tree). Next I would add a small piece of code to my IParse program to read s-expression and use the unparse abilities with the C grammar to produce nicely formatted C code. The unparse function is smart enough to output a S-expression like '(times (add 3 4) 5) into (3+4)*5 based on the given grammar rules fed to IParse. Feel free to contact me, if you want any help with using IParse.
User Journal

Journal Journal: Kapla: Spirals

Just building stuff, sometimes continuing the following week: Week 1 Week 2

The booklet had a spiral which took some time to figure out. Basically, 2 pieces in the middle, and 1 on each side of it, which makes it look really cool, and more importantly, supports the turned pieces. They do a slow turn though.

Comment Interpretting instead of compiling (Score 2) 674

I once discovered that an interpretting parser was faster than a compiled version of it. Probably because the compiled version became so large that it was larger than the CPU caches, causing lots of reads from memory, while the interpretter did not. It is often the case that interpretting is fast enough and that there is no need for compiling (to machine code or virtual machine).

Comment Re:wireless sensor network (Score 1) 67

The LoRa network has some measurements against these kind of attacks. I built a LoRa node myself (to connect to the free network owned by TheThingsNetwork.org with only limited coverage, see ttnmapper.org) and had to generate two private/public keys, one for the wireless connection and one for the application, meaning that the transmitted information is encrypted. There is also frame counting implemented. The maximum rate on this free network is limited to about 80 bytes per hour. Maybe the commercial network has a slightly higher rate and is bi-directional. The railway station of Utrecht has connections in five directions and is know to cause massive delays in the very tightly scedule rail network in the Nethelands due to (for example) frozen rail switches. So, I guess they are only using it to signal these kind of things. Not something as vital as the position of the switch.

Comment Self taught (Score 1) 515

I learned Algol 60, FORTRAN, and Lisp while still in high school. Writing programs out on paper before traveling to a near by city by train to make punch cards and run the programs, often going home with s stack of listings to study the results. Only in second year of university did I learn something truely new: currying and lazy evaluation. More details.

Comment Re:It time, this will be good for Lee Sedol (Score 1) 111

I felt that both the commentaries on the official stream and the one of AYA with Kim Myungwan were mostly focusing on analysing alternatives to moves that were already played, instead of focusing on possible moves to be played. At one point Haijin Lee, also known as Haylee, the current secretary of the International Go Federation, took a seat behind the computer. While Lee Sedol was taking a long time to think about the 78 move, she suggested the "wedge" move to Kim Myungwan. He first asked with some surpise: "This one?", and after she confirmed he exclaimed: "O, that is very creative". They analyzed the various possibilities to take benefit of the move, but did not see how it could work. Haijin at one time remarked, "It has to work", indeed because otherwise the game would be a loss. Then with still 6:16 on the clock, Lee Sedol played the proposed "wedge" move at L11. Then Haylee has to go. Then AlphaGo plays the move that Kim Myungwan just played as a possible continuation. Soon it is acknowledged that it is a mistake and that the chances for Lee Sedol have turned. The "wedge" move is later recognized as the genius move with which Lee Sedol defeats AlphaGo. Demis Hassabis tweeted Mistake was on move 79, but #AlphaGo only came to that realisation on around move 87 and When I say 'thought' and 'realisation' I just mean the output of #AlphaGo value net. It was around 70% at move 79 and then dived on move 87. The question remains if there is another move that AlphaGo could have played to prevent Lee Sedol from winning. If this is not the case, then move 78 is simply an over play that tricked AlphaGo.

It seems that AlphaGo suffers from the horizon effect, meaning that when it is ahead, it becomes blind for wining sequence that starts with an unusual move. It is great that Lee Sedol has won this game, because it shows that humans still have a better understanding of the game. Relying on a mental vocabular for reasoning about the game, has some weaknesses, it might still prove superior when reasoning about very complex situations like the one encountered in this game. It seems that Lee Sedol took most time to think about this move. Later on in the game, AlphaGo, played some bad moves, probably because its winning percentage had dropped.

Slashdot Top Deals

As in certain cults it is possible to kill a process if you know its true name. -- Ken Thompson and Dennis M. Ritchie

Working...