МОСКОВСКИЙ ГОСУДАРСТВЕННЫЙ ТЕХНИЧЕСКИЙ УНИВЕРСИТЕТ имени Н. Э. БАУМАНА

Е. А. Яковлева

МЕТОДИЧЕСКИЕ УКАЗАНИЯ

По обучению чтению на английском языке научно-технической литературы по специальности

«Автономные информационные и управляемые системы»

Москва

2005

Unit 1

I. New words and expressions

1. boom (n.) - гул, рокот, звуковой удар, быстрое развитие

2. bridge (v.) - преодолевать препятствие

3. assertion (n.) - утверждение

4. predominant (adj.) - преобладающий

5. query (n.) - вопрос

6. core (n.) - сущность, ядро

7. neural (adj.) - нервный (мед.)

8. gap (n.) - брешь, разрыв

9. cognitive (adj.) - познавательный

10. launch (v.) - начинать, запускать

11. consciousness (n.) - сознание

12. domain (n.) - область, сфера

13. reasoning (n.) - а. логическое мышление; б. доводы

14. approach (n.) - подход, приближение, метод

15. impose (v.) - навязать, возлагать

16. establish (v.) - основать, организовать

17. ingenious (adj.) - изобретательный, искусный, оригинальный

18. claim (v.) - требовать, претендовать, утверждать

19. magnitude (n.) - величина

20. feasibility (n.) - возможность

21. facilities (n.) - средства, устройства, аппаратура

22. generality (n.) - утверждения общего характера

23. adoptable (adj.) - приемлемый

24. adaptability (n.) - адаптивность

25. disregard (v.) - игнорировать, пренебрегать

26. tie (v.) - связывать, соединять

27. execute (v.) - выполнять

II. Match Russian and English equivalents

1. artificial intelligence 1. навязать свой взгляд

2. to produce a boom 2. сравнение доводов «за» и

«против»

3. to bridge the gap 3. широкий выбор логических

устройств

4. to launch a project 4. создать системы

5. to give on-line answers 5. заявить права на

6. a wide range of reasoning facilities 6. ликвидировать разрыв

7. comparison of pros and cons (лат.) 7. дать ответ в режиме реального

времени

8. order of magnitude 8. искусственный интеллект

9. to adopt an approach 9. обеспечить быстрый рост

10. to claim one’s rights 10. принять научный метод

11. to establish a system 11. полагаться на односторонний

подход

12. to rely on one-sided approach 12. начать работу над проектом

13. to impose its own view 13. порядок величины

III. Translate the following pairs of sentences. Mind different contextual meanings of the italicized words.

1. The artificial satellite launched was manufactured in Russia. The fifth project was launched in Japan. A launch of a space vehicle is a complicated process.

2. Ferrous metals are subject to corrosion. Three subjects were predominant in making a copy of a human brain. Artificial intelligence is probably the subject to challenge human interest forever. Data were subjected to thorough verification.

3. Reasoning is done through argumentation, comparison and pro and con arguments. The power of reasoning belongs to “homo sapiens”.

4. Both projects have adopted the computationally strong-AI approach. Tsiolkovsky’s ideas were adopted by his followers.

5. AI technology is undergoing a boom. Aircraft manufacturers strive to make the boom of aircraft lower.

6. To bridge a gap between the two theories scientists worked more than a decade. The arches of the bridge were high enough to let a tall ship pass.

V. Translate the following derivatives stating the part of speech they belong to

Regard, disregard, regardless, general, generality, adopt, adoption, adoptability, conscious, consciousness, unconscious, intend, intention, intentionality, unintended, execute, executive, execution

V. Read and translate the text Text 1a

ARTIFICIAL INTELLIGENCE

To understand current trends in artificial intelligence, the history of AI can be of great help.

The first AI era was impressed by the fact that human brains are several orders of magnitude slower than computers (in transmission as well as coupling speed). Therefore, making a copy of a human brain on a computer would have to result in something ingeniously better. Three subjects were predominant: (1) learning without knowledge, (2) neural modeling (self organizing systems and decision space techniques), and (3) evolutionary learning.

Major AI projects have not resulted in intelligent or commercially successful products. But intelligence cannot be easily achieved on digital computers with existing approaches. Today’s computers as well as existing approaches basically do not differ much from those of 30 years ago (apart from being faster and having better storing capacities) and, therefore, are very unlikely to approach not only human level but also any level of intelligence established by biological intelligent systems.

Several other projects were started, based on logic programming (LP). The project was heavily based on logic programming to bridge the gap between applications and machines.

Japan has already launched the Sixth Generation project, based on real-life domains, neural networks, optical connections, and heavy parallelism. This project addresses the tremendous task of codifying a vast quantity of knowledge possessed by a typical human into a workable system. General assertions have been put into CYC’s knowledge base, using a vocabulary with approximately 10³ atomic terms. CYC is intended to be able to give on-line sensible answers to all sensible queries. CYC includes a wide range of reasoning facilities, including general deduction and analogical inference. Reasoning is done through argumentation, through comparison of pros and cons.

CYC is the first project of this magnitude. All knowledge in CYC is encoded in the form of logical sentences, and not in diagrams, procedures, semantic nets, or neural networks.

The two projects mentioned could not produce intelligent systems at all. They have adopted the computationally strong AI approach instead of at least combining it with others, e.g. cognitive weak-AI. Both projects relied on a one-sided approach, disregarding the new school of AI. This new approach claims that to design an intelligent system, none has to give it all properties of intelligent creatures: unity (i.e. multiple knowledge and multistrategy approach), intentionality, consciousness and autonomy along with generality and adaptability.

One algorithm executed on a single processor cannot emulate intelligence. The process must consist of many interleaving and intensively communicating sub processes. The general approach seems promising, yet it is not clear in which particular direction the discovery of true intelligence lies. For the time being it seems that new AI is strongly related to interdisciplinary sciences, especially biological and cognitive sciences.

The strong thesis of multiple knowledge states that multiple semantic models are an integral and necessary part of intelligence in any machine or being in real-life domains. A single model cannot achieve as good performance as multiple models because each model ties to fit data and noise according to its own structure and therefore ties to impose its own view. In other words, although multiple models can be at any time transformed into one single model with the same performance as a set of models, in general it is not possible to construct such a single model in the process of learning without designing multiple models.

Integration of models after they are designed seems not only feasible but also sensible because of reduction in storage and classification time.

99% of all existing computer systems and most current AI orientations are based of a single model. Intelligent systems seem to have special properties, e.g. multiplicity.

(3500 characters)