15

Common sense knowledge is the collection of premises that everyone, in a certain context (hence common sense knowledge might be a function of the context), takes for granted. There would exist a lot of miscommunication between a human and an AI if the AI did not possess common sense knowledge. Therefore, common sense knowledge is fundamental to human-AI ...


10

We need this kind of common sense knowledge if we want to get computers to understand human language. It's easy for a computer program to analyse the grammatical structure of the example you give, but in order to understand its meaning we need to know the possible contexts, which is what you refer to as "common sense" here. This was emphasised a lot in ...


5

Yes, it is possible to combine probabilistic / bayesian reasoning and a traditional "knowledgebase". And some work along those lines has been done. See, for example, ProbLog ("Probabilistic Prolog") which combines logic programming and probabilistic elements. See: https://dtai.cs.kuleuven.be/problog/tutorial/mpe/01_bn.html Another project to look at ...


4

First of all, I would like to point out the main differences between knowledge base and (Deep) machine learning, specially when the main focus is on "AI" not "Data Science": NNs are like a black box; Even if they learn a dataset and gain the power of generalization over the problem domain, you'd never know how they are working. if you scrutinize the details ...


4

Neural nets incorporate prior knowledge. This can be done in two ways: the first (most frequent and more robust) is in data augmentation. For example in convolutional networks, if we know that the "value" (whatever that is, class/regression) of the object we are looking is rotational/translational invariant (our prior knowledge), then we augment the data ...


4

I want to preface this by saying that the distinction is not clear. Nevertheless, I'll tell you what I know about this, and I will attempt to make the further clarification: The Structure of rule-based agents is: Take input from environment, pass through condition-based rules, and perform the action through actuators or anything which creates some ...


4

I find it unlikely that you'll find a firm answer, so I will try my best to guide you towards information which may help you form an opinion either way. Turing had the controversial opinion (which remains controversial today) that: Digital computers have often been described as mechanical brains. Most scientists probably regard this description as a ...


4

Although it is common to represent a grid as two dimensional array in a computer program, this is not the only way to represent one. You could, for example, use a generalized graph structure made of linked nodes, with 4 links each. Many other representations are possible. Even if you use a 2d array, some languages would index the columns first rather than ...


3

I will first recapitulate the key concepts which you need to know in order to understand the answer to your question (which will be very simple, because I will just try to clarify what is given as a "definition"). In logic, a formula is e.g. $f$, $\lnot f$, $f \land g$, where $f$ can be e.g. the proposition (or variable) "today it will rain". So, in a (...


3

To my knowledge, this is very much an open research issue. Here is a paper by Prof Leslie Smith, an acknowledged expert on neuromorphic perceptual coding, which explains the importance of the notion of perceptual time for Artificial General Intelligence and sketches an architecture from which a notion of 'now' might emerge: Perceptual Time.


3

Your question depends heavily on the method you are using for machine learning. It sounds like you want to extract certain features like "curves and straight lines" from your images and use them as training data. This step of extraction is usually not considered part of the training process but part of pre-processing. During pre-precessing you read your ...


3

Yes, we can do it in a deep learner. For example, suppose we have an input vector likes (a, b) and from prior knowledge, we know a^2 + b^2 is important too. Hence, we can add this value to the vectors likes (a, b, a^2 + b^2). As another example, suppose date time is important in your data, but not encoded in the input vector. We can add this to the input ...


3

StarCraft II is a real time strategy game that combines fast paced micro actions with the need for high level planning and execution. StarCraft II being a popular game with millions of users it proceeds that defeating top players becomes a meaningful and measurable long term objective in AI research. Computer games provide a compelling solution to the ...


3

In the first sentence of the paper, Nilsson states that [s]everal artificial intelligence (AI) applications require the ability to reason with uncertain information. Nothing (well, almost nothing) is ever just true or false, and binary logic is not enough to model a complex world. So we need more powerful means of describing logical relationships that go ...


2

Theory of mind Relationships and normal social behavior require a human to possess a reasonable "theory of mind", a skill in understanding and modeling the thought processes that happen in the minds of others, and making reasonably accurate predictions on how particular actions will be understood by others. In general, this might be treated as any other ...


2

In general, what you are describing implies a hierarchical sequence model, in which mannerisms adapt to the regime or paradigm in effect. Expressive modalities are how we recognize the operative context from the behaviors of other agents. For an artificial agent, avoiding un-canniness would involve clustering the factors underlying the classification of ...


2

To add to the Foivos's answer, Convolutional Neural Networks are shift-invariant. Fukushima introduced this to his Neocognitron. There is a trail to introduce scale-invariance to CNN. https://arxiv.org/abs/1411.6369 Also, CNN uses structural characteristic for the prior knowledge. And neural networks are locally smooth. It is not perfect, but neural ...


2

Although I gave an up vote because the question, in principle, is a good one, there is considerable technology ground to cover before such a project would be feasible. A century or two might not be enough time to cover the ground, although no one knows. All past guesses have covered the range from a decade (which already passed, so that was grossly ...


2

Some of the work on descriptive logics and modal logics was done within the context of artificial intelligence from a research funding perspective. Some was part of the normal academic apparatus of mathematics departments. Furtherance of these fields in the AI context has been hindered by historically low return on investment. Although first order logic, ...


2

First of all, sigmoid does not output 0 or 1, it outputs any real number in the range between 0 and 1. Furthermore, normal neural networks doesn't output binary values, unless the output layer uses the step function as activation function (which is rare). I'm not really sure if you want the NN to be a classifier or regressor, but it sounds like you want a ...


2

I'll answer this question in several parts: Why do AGI systems need to have common sense? Humans in the wild reason and communicate using common sense more than they do with strict logic, you can see this by noting that it is easier to appeal to someone's emotion than logic. So any system that seeks to replicate human cognition (as in AGI) should also ...


1

Obviously you will have to cover a lot of technological ground before you start thinking about actually starting work. I may be wrong but here is what i think could be one of the solutions what areas you can focus on to begin work. At least theoretically. Step 1: Teach the bot to interpret text,it should be able to distinguish between words, sentences. ...


1

In the context of case-based reasoning the chunks are called “cases” and are equal to memory from the past. Let us investigate an example how a knowledge based expert system looks like. At first, we have the knowledge itself which is determined by symbolic events happening over time: time event 0 carIsArriving 1 personIsLeavingCar 1 carIsArriving ...


1

It kinda depends on how exactly you define knowledge, and what you believe about what the weights in a trained NN model really represent. But to answer this question in the most straightforward possible way (hopefully without sounding glib), then yes, a NN can be pre-trained, and then you can take that model and apply additional training to it, so in a sense,...


1

The use of knowledge representation and planning to create and improve an algorithm is an area of intense interest. Parallel and Distributed Processing In a technology culture where parallelism is of even greater interest, we see an effort to automate the creation of processes at higher levels in multiple approaches to improve speed and reduce the cost of ...


1

So called automatic programming is used to generate sourcecode from abstract descriptions. The input is usually a requirement specification and the aim of the planner is to find the executable code. In the literature this concept is described as Algorithm prototyping. Notable examples are Gocad and MAPP (the Berkeley Model and Algorithm Prototyping Platform)....


1

I think Demento has answered it well, but probably below can add some more understanding to you. [1] Usually the data stored for image processing relates to the Image Pixel Densities(as per the many courses i visited online), which can be very well maintained using a matrix of pixel density values, corresponding to the resolution of each image. Then it ...


1

Rule-based systems cover a wide range of systems. Some make use of boolean if/then/else rules, others may use weighting or even probabilistic inference. Some operate on frames, some on java objects, some on propositions that can be formulated in predicate logic. An example of a popular rule system is Drools. Some rule systems can be expressed as a subset of ...


1

It seems that they are stating that a knowledge base is consistent if and only if it never asserts the truth of both the truth and the negation of a particular P. In other words, a knowledge base is consistent if it never contradicts itself. Their definition allows incomplete knowledge bases to be considered consistent; by their definition, an empty ...


1

A very large one, the world wide web with highly scaled and optimized indexing by Google.com is the most distributed and robust schema-agnostic database known today. Without the schema-awareness Google brought to the table by applying more rigorous information science to the table, it was almost useless to those that did not know the URL of the target ...


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