RSS
 

Arquivo para a ‘Inteligência Artificial’ Categoria

Theory of Information or Signals

07 May

David Hilbert proposed 23 problems for Mathematics to solve, among them the 2nd. problem was the consistency of Arithmetic to solve problems (which started the idea of algorithm). Later from 1910 to 1913 Alfred Whitehead and Bertrand Russerl published the 3 volumes of Principia Mathematica recalling Isaac Newton’s Principia (Philosophiae naturalis principia mathematica) of 1980, but logic has shown contradictions.
It was Kurt Gödel (1906-1978) who showed the contradictions of logic by demonstrating that a mathematical system is either complete or consistent, a problem that was discussed by Alan Turing and Claude Shannon at Bell Labs in the late 1930s amidst World War II .
Both were in secret projects, Turing was to decipher the Enigma machine, captured from the German army, and Shannon build the System X machine to encode messages from Roosevelt and Churchill from the allied troops.
Both spoke at the lab meetings about Gödel’s problem, and wondered if the machine could think. Boole’s circuits and the Turing machine are at the origin of modern computers, this information theory is actually information in another context, Shannon himself would call it a message, and modern computing calls it signals.
The computation evolved with the construction of the first computer by von Neuman, and we initiate at the moment to ask again if the machine will surpass the human, the philosopher Sam Harris has a opinion:

<

 

AI will help fight fires

21 Feb

The famous US Department of Defense (DoD), which is also on the internet, has launched a program to use artificial intelligence (AI) for drone data analysis to improve how forest attacks are fought, the news is on Wall Street Journal this week.
The project uses algorithms that evaluate photos and videos, being themselves the ways of forest fuels and improving efforts to contain them, data on resources in helicopters, but also using drones and other cyberspace data on the administrative functions of as part of efforts to improve its efficiency.
A program to monitor and attack forests in California and other parts of the country is one of two pilot projects the Pentagon released on Tuesday.
The program, solved the Pentagon, oriented strategies to establish new innovation strategy of artificial govern, that must to work performance with the academy and the industry for better to advanced information management and data application.
The reason for the fires is clear, also in the Portuguese rhodians, which are a combination of vulnerabilities in the work plans of children can issue fires, such as the serious happened at Camp Fire in 2018 in California, see CNBC vídeo:

 

Even waves can go to AI

20 Feb

The OpenAI project, although it is said to be “non-profit” and is really open, just enter the blog of the project to check the progress and the possibilities, in fact even pointed the experts when publishing a system that writes news / texts, theoretically fictions, but that can be classified as fake, or as they are being called: faketexts.  Also the code is open and available on the GitHub site for developers. 
Natural language processing systems can perform tasks such as answering questions, machine conversion, reading comprehension, and text summarization, which are already typically addressed with supervised learning in task-specific data sets, but text search in GPT2 is wider in quantity.
GPT2, the successor to the GPT that was just a producer of texts from basic texts, can now read up to 40 GB of existing text on the Web, and what it has produced frightens a little, for clarity, depth and worst of all, pure fiction or more clearly: fakes.
Among its syntactic characteristics, it is superior to others of the sort, writing passages that make sense with the previous text and maintains the style, without getting lost in long sentences. The problem is that it can generate fakenews that can now be longer, becoming faketexts, a report from The Guardian shows the novelty and the problems:

 

Markham in Canada: smart city

13 Feb

Ignorance of the benefits that smart and rational use can bring to many cities make many public policies inefficient and unnecessarily expensive, a serious problem is the infrastructure to monitor cities. In a partnership project, Bell Canada and IBM Canada want to make the city of Markham,
Ontario in Canada, with approximately 330,000 inhabitants, a model for using networks to make it a “smart city,” according to the Financial Post
The city government will use Bell’s broadband networks and IBM data analysis, collecting sensor data placed throughout the city to decide on investments in various aspects of infrastructure: pipeline leaks, flood-prone regions (Lake Ontario), problems that arise in storms, effective use of energy and tracking.
The projects do not fail to include social improvements and ecological aspects that also depend on the good use of technology and energy,
Many environmentalists are unaware that the results of Medellin were also achieved using Smart Cities concepts, being the first case in Latin America.
South. Similar trials and projects are being conducted in Canada and other parts of the world in view of which a number of multi-provider services may be preparing services that best cost and prevent service disruption.

 

Wikipedia and Artificial Intelligence

24 Oct

Having already almost surpassed the point of singularity (see our post), the point that the machine would surpass human intelligence, the question now turns to consciousness and a well-considered point is the question of consciousness.
In this sense the main criticism is the perpetuation of prejudices, which would avoid what I call hermeneutics, but it is an incorrect view of the evolution of digital technology, for example, the use of Digital Ontologies and the ability to seek scientific studies outside of Wikipedia.
This is what recently announced an article in The Verge, and the most serious omission after researching scientists who are omitted from Wikipedia, was to note that 82% of written biographies are about men.
In a blog post, according to The Verge website, John Bohannon, director of science at Primer, explains the development of Quicksilver tool to read 500 million original documents, sift through the most cited numbers and then write a basic article about the work of these scientists not mentioned in Wikipedia.
Two examples of illustrious women found and written for AI are Teresa Woodruff, a scientist who designed mice ovaries using 3D printers, was cited by Time magazine in 2013 with one of the most influential people in the world scientist, and the other case is that of Jessica Wade, a physicist at Imperial College London, who wrote the new entry for Pineau.
Wade was one of the scientists who said “Wikipedia is incredibly tantalizing, and the underrepresentation of women in science is particularly bad,” and praised Quicksilver stating that with it you can quickly find large amounts of information very quickly.
Wikipedia will have to evolve with Machine Learning tool, this may happen in the coming years, the fact that there are specific tools for this does not invalidate Wikipedia, shows that it has weaknesses and should be corrected

 

The paraconsistent logic

25 Sep

Kurt Gödel’s paradox, that a complete system is inconsistent was fundamental to a new phase in formal logical principles, and cooperated with the emergence of the computer.

It was the Peruvian philosopher Francisco Miró Quesada, unknown to many Latin American scholars, who coined the word paraconsistent in 1976.

The Brazilian Newton da Costa who developed this theory has become very important in several areas, including philosophy and Artificial Intelligence. In the figure, the axis to the side that goes from u 0  and u  1 ,  is called degree of belief, but the truth has points A for consistent and C for inconsistent, having a lot of application to everyday life.

The study applied to semantics mainly explores paradoxes, for example, it can be said that a blind man sees certain circumstances, also the study of different forms of perception and abilities can help the parameters of the deep mind.

We have already stated that neologicism can precisely help AI in this phase of deep intelligence, for example, in the study of natural languages, the language of everyday life.

The idea that A and non-A could exist was inconceivable in Western philosophy, is the principle of the excluded third, which comes from Parmenides and was consolidated in Aristotle. Paraconsistent logics are purposively more “weak,” terms to refer to this break with classical logic, for they solve few valid propositional inferences in the classical sense of logic The logic of paraconsistent languages, however, are more conservative than those of classical counterparts, and this changes Alfred Tarski‘s hierarchy of metalanguage.

The influence on natural language was anticipated in 1984 by Solomon Feferman who stated “… natural language abounds in expressions directly or indirectly self-referential, although seemingly harmless, all of which are excluded from the, because in daily life, in truth we are paraconsistentes

 

And you who say that I am

21 Sep

 The truths of the facts only reveal in the truth of the acts, it is so for the daily life, so for the politics and for the speeches, if we live in the post-truth, it has the limit of the acts.
We like day to day to create narratives more favorable to our convenience and our ideals, but almost always the unveiling exists beyond language and speech.
The creation of a logical intelligence, in deeper layers now called Deep Mind or Deep Learning, is nothing more than the artificial response to the virtual world, part of the real next to the current one, a logic consistent with the action, the human will always be some dyslexia.
Full consciousness is linked to full dialogue, where discourses can interpenetrate in the hermeneutic circle, the difference with artificial intelligence is that the machine learns from humans, but it will be difficult for it to escape from formal logic, while the human is ontological.
This means that we are in the age of Being, a deeper manifestation of what we are, and contrary to what anti-technology discourse supposes, it is precisely this that can help human speech in the essential aspects of logic, which we sometimes falsify to be correct .
Historically technology is not displaced from human needs, it is often the poor adaptation or use of human relationship with technology that causes some disruption and misunderstanding of its true role, which is to assist the craft, art and technique, says the Greek origin of the word techné.
In the biblical passage that Jesus tests his apostles, he asks: “And you who say that I am” (Mt 16:15), for some was a great prophet, for others a return from Elijah or even from Moses, not always God , that is, the Divine wisdom between us.
The reprehensible use of the evangelical message in politics is not for the fact that they should be outside the interests of the common good and of society in general, but it is the possibility of instrumentalizing and using in favor of a certain discourse that is not always coherent with the gospel

 

Is it possible measure the evolution of AI ?

30 Jul

Now it seems that yes, it is necessary to establish more precise tests and exactly the accuracy of these tests, but researchers of the Computer Vision Lab, of the Federal Institute of Technology of Zurich have created a test system (benchmark) that tests the performance of neural network´s platforms used to perform common Artificial Intelligence tasks, such as answering questions, finding certain data, and other smartphone applications already used in common neural network tasks.

The idea is to measure the performance of AI systems, such as and done on today’s computers and on applications useful in smartphones, which use Artificial Intelligence.

The basic plan + and measure the performance in the systems, as already done for some answers: time, locators, flights, restaurants, etc. and growing in complexity, which will make manufacturers and application vendors make their models more sophisticated.

The application calls AI Benchmark and allows, among other things, the comparison of the speed of AI models to run on different Android smartphones, punctuating performance, already found in the Google Play app.

The device that has the dedicated AI chip dedicated to some devices (at least with Android OS 4.1 installed already will be fast enough to do these functions.

Some applications may be more complex, for example sorting images with easy recognition and the ability to segment and improve photos, indicating that this type of application can grow and evolve to more complex applications.

 

AI learning with animals

03 Jul

Beavers, termites and other creatures build structures in response to environmental problems, the idea of ​​using these strategies in autonomous robots was made by researchers at the University of Buffalo. In the new system the robot continuously monitors and modifies its terrain to make it more mobile, similar to the beaver that reacts to the water flow building a dam, what kind of intelligence is this, Thomas Nagel would ask what it’s like to be a beaver?

The problem is not so simple and the algorithms must change when there are unpredictable and complex spaces, for this resorted to a biological phenomenon called stigmatics, which is an indirect co-ordination reacting to a problem, biologists and zoologists study this. Researchers using this new algorithm, equipping the robot with a camera, a specialized software and a robotic arm that lifts and deposits objects, placed bean bags of different sizes around the area, in 10 tests, the robot changed from 33 to 170 bags, each time creating new ramps to reach their destination.

A release of this work is on the University of Buffalo website, and the paper will be presented last week (June 25-30) at the Robots: Science and Systems conference in Pittsburgh.