# Tag Info

18

There is no "official Turing test" so there's no concept of "officially pass[ing] the test". Turing described a methodology that one might use to evaluate artificial intelligences. The organizers of the event that Eugene Goostman won implemented that methodology in a particular way and the program satisfied the criteria the organizers had chosen. In that ...

18

The Church-Turing thesis says that the informal notion of an algorithm as a sequence of instructions coincides with Turing machines. Equivalently, it says that any reasonable model of computation has the same power as Turing machines. An artificial intelligence is a computer program, i.e., an algorithm. If the Church-Turing thesis holds, then you could ...

18

Recoverable, as we can pour all water back to the 12L jug to restore the original state, hence any state derived thereof (by following the same steps from the start). The problem is, also, solvable: [12, 0, 0] [4, 8, 0] [4, 3, 5] [9, 3, 0] [9, 0, 3] [1, 8, 3] [1, 6, 5] [6, 6, 0] # <-- SOLVED; 7 steps DETAILS: The problem stated as-is is ...

17

The halting problem is not a statement about intelligence (human or artificial) it is a statement about the limits of mathematics. It is an historically important example of an undecidable problem. An artificial (or human) intelligence can certainly look for and find many sorts of different infinite loops in real programs. And the halting problem doesn't ...

16

The whole state space for chess is enormous - it can be roughly estimated as 1043 (Shannon number (Shannon, 1950), (Wikipedia)). The idea you present - Reinforcement Learning agents playing with each other to learn the game - was successfully applied to Backgammon - TD-Gammon (Tesauro, 1995), (Chapter in Reinforcement Learning by Sutton&Barto). It also ...

10

Although Alan Turing is of course a very important computer scientist, the Turing test is only superficially related to computer science. It is more related to philosophy. As long as machines exist people have wondered whether it is possible to construct a machine that can think. But to answer this question we must first answer the question what it means -- ...

9

The most common/obvious way is a challenge-response test that is easy for humans but hard for computers (of course, but not only, CAPTCHA). This kind of test is very effective{1} but falls under the HIP (Human Interactive Proofs) area: it's not transparent. Typical, "simple" approaches to distinguish human website traffic from Bot are: time it takes to ...

9

Roughly speaking, over-fitting typically occurs when the ratio $\frac{\text{complexity of the model}}{\text{training set size}}$ is too high. Think of over-fitting as a situation where your model learn the training data by heart instead of learning the big pictures which prevent it from being able to generalized to the test data: this happens when the model ...

9

I think the prizes you're referring to are the Loebner Prize. According to the Wikipedia page (see prior link), the winner for 2014 is 'Rose' by Bruce Wilcox. That program did not win one of the one-time-only prizes, but did get $4,000 in prize money. 'Eugene Goostman' competed in 2005 and 2008, finishing second both times. The competition 'Eugene Goostman'... 8 Connect Four was solved in 1988. The first solution was given by Allen and, in the same year, Allis coded VICTOR which actually won the computer-game olympiad in the category of connect four. I would suggest you to go to Victor Allis' PhD who graduated in September 1994. You can get a copy of his PhD here. In Section 6.3.2 Connect-Four (page 163) you can ... 8 Turing machines are a model of computation, one way of formally defining the concept of an algorithm. While Turing machines are usually defined using barebones input/output capabilities, it is not difficult to extend them so that they can apply to the kind of problems that AI is after. The relation between AI and Turing machines is thus: Artificial ... 7 The deterministic requirement isn't all that constraining. That just implies your vehicle is certain of the state it's in. That being said, you'll probably want to plan paths in a way that allows you to avoid obstacles. The best way I've seen this done is with sampling-based planners. Steven LaValle wrote the central academic resource on this topic: Planning ... 7 At a stretch it is an expert system (such as fuzzy logic). As you are not running an algorithm to perform feedback onto the decision parameters based on the output, it's not really learning. However, performing feedback is not the only indicator whether an alogirthm is AI. One could argue that if it acts in a way that appears intelligent, that's all that ... 7 The idea of DeepDream is this: pick some layer from the network (usually a convolutional layer), pass the starting image through the network to extract features at the chosen layer, set the gradient at that layer equal to the activations themselves, and then backpropagate to the image. Why does it make sense? Intuitively, it amplifies the features that are ... 7 Absolutely yes! your arguments are correct. And, as matter of fact, it is very easy to come up with a graph where Bidirectional Dijkstra would expand more nodes than Unidirectional Dijkstra, following your reasoning. Consider a graph$G(V, E)$where$V$is the set of vertices,$E$is the set of edges and you want to find the shortest path between two ... 7 You're understanding of "solve all instances of the Halting Problem" is flawed. All instances means ALL instances. Every single one, ever. For example, no human knows if a Turing Machine verifying the Collatz Conjecture will halt. Bam, right there, your premise that humans can solve "all" incarnations of the Halting problem is flawed. There is no reason to ... 7 I couldn't in a quick Google through the textbooks tell whether recoverability requires immediate, single-move recoverability (ctrl-Z-style) with each undo step being O(1) complexity, and which can be implementable as a sequence of moves in a push/pop stack. They always use guaranteed-single-move-undo examples (8-puzzle, Towers of Hanoi) for their ... 6 Without any a priori knowledge, the problem is insoluble. There are infinitely many possible answers and you have no wat at all to say that one of them is preferable to any of the others. How can you possibly tell just by manipulating symbols that the sequence is "Counting in binary using A and B for 0 and 1" rather than "B, BA, BB, BAA, BAB, BBA, BBB, BAAA, ... 6 Reinforcement Learning (RL) algorithms are useful in a setting when the structure of the environment is unknown and/or stochastic. The environment is normally formalized as a Markov Decision Process (MDP) or Partially Observable Markov Decision Process (POMDP) and the goal of an RL agent is then to maximize its cumulative reward (or minimize its regret with ... 6 A Turing machine is a tuple consisting of an alphabet, a tape, some transition, etc. I am not one of those. So no. A human being is not a Turing machine (or at least, I am not). 6 Many implementations you can find out in the web are done on matrices (MATLAB for instance) since it provides a compact notation. Haykin's textbook on neural networks takes this approach. Matrices also provide a simple translation to hardware design (FPGA, ASIC, etc.). They are also more often implemented on the FPU. If you implement a neural network in an ... 6 A brief definition would be: Cognitive computing is the simulation of human thought processes in a computerized model. more detailed explanation: Cognitive computing involves self-learning systems that use data mining, pattern recognition and natural language processing to mimic the way the human brain works. The goal of cognitive computing is to ... 6 I don't know about robotics, but ontologies are part of the standard toolkit for modern expert systems, especially those with a natural language processing component. For example, consider the process of performing literature searches for systematic reviews in medicine. Of the millions of medical studies out there, reviewers need to find the 20 or so high-... 6 Well, as far as I can tell, there was no specific automated planning tool for the Rosetta mission as we understand automated planning systems in ICAPS, which is the most prominent conference on Automated Planning and Scheduling. The hardest part was to plan the journey from earth to a common location where both the spacecraft and the comet could meet and ... 6 The discount factor does not represent the likelihood to reach the state$s′ $from the state$s$. That would be$p(s'|s,a)$, which is not used in Q-Learning, since it is model-free (only model-based reinforcement learning methods use those transition probabilities). The discount factor$γ$is a hyperparameter tuned by the user which represents how much ... 6 Let me please start by succintly summarizing the behaviour of RBFS. For a thorough explanation of the algorithm refer to the original journal paper: Richard Korf. Linear-space best-first search. Artificial Intelligence, 62 (41--78), 1993. In fact, RBFS is much more than "just updating the$F(n)\$ value for the parent for which the current execution was ...

6

With an admissible heuristic The heuristic defines which nodes will be explored first, but does not change the final path found. In your example, the heuristic will cause the path to Z to be explored first. The algorithm will discover the true (expensive) cost of 120. Then it will decide to explore the A path and discover the optimal route. A better ...

6

These are known as Autoencoders. As you said, these neural networks are trained to produce output that is similar to the input, rather than output a classification of some kind. Internally, they do not differ much from other neural net designs. Simply, the expected output is the input (or a slight variant of the input), rather than a classification. One ...

6

I was wondering the same and just stumbled across a nice tutorial by Quoc V. Le. The following explanation deals with the conditional case since this seems to be the common case. My explanation is based on and the image is taken from chapter 5 Sequence output prediction with Recurrent Neural Networks. Background We only regard a decoder with a single cell ...

6

Even if we can design better AI solving problems we did not think possible even 10 years ago, they will always be computer programs. And as you said it yourself, "Alan Turing states that, there can't be a program that can decide if a program will ever stop." (he actually proved it and not only stated it...). Thus even the best AI you can think of will not be ...

Only top voted, non community-wiki answers of a minimum length are eligible