1 year, 12 months ago
This article is devoted to research of opportunities of neural networks at their use as basis for individual reason of modeled object.
Purpose: to show, whether the neural network is capable (or its this implementation) to perceive the "surrounding" world, independently to be trained and on the basis of own experience to make decisions which can be considered rather reasonable.
1 year, 12 months ago
If to walk on colleagues and to ask how many them cell phones, it will appear that on average them about 2.5, but thus at the vast majority their no more than one. There is at once set of questions beginning from, why their suddenly not integral number and how after all to evaluate how many phones on average at the person.
The median assessment will be suitable for such purposes. That is such statistics that half of values of selection is less, and half is more. More formally: let's arrange values of selection on order and we will select among them with sequence number. Such assessment has some advantages. It is less subject to influence of wrong data, value will always be from that set that met in selection, but there are also unpleasant shortcomings, main of them, it is complexity of calculation, even for enough widespread distributions there is no general calculation formula (is more precisely, but it is difficult to put it into practice, watch Distribution of serial statistics).
2 years ago
Remained very little before the termination of set in the third School of interface developers which this time will pass in Moscow. Emphasis in it will be put on practice in format look mini-hakatonov. We have already tested it last year in Minsk and Yekaterinburg. Students will share on teams, and is already command to implement the project. Except writing of the code, it will be necessary to be able to make decisions, to deal with the arisen controversial issues, to break all development process into logical iterations. Children from Yandex who will individually work with each team will help with it. Occupations will begin on September 7.
Three years our colleagues make tasks, check works and give lectures within School. Besides, they face every day real working tasks, being engaged in development of interfaces of services which visit millions of people with different devices and different browsers.
Entrance test in School too the very practical. We will stop accepting requests on August 16 at 23:59. While there is still time to cope with tasks, we have asked our teachers to help a little future students and on the example of the last year's questionnaire to explain, by what logic it is worth being guided, solving the offered problems, and to tell to that they pay attention at their check.
This year the structure of tasks is constructed by analogy with offered last year — so that we could pay attention to different aspects of development. The first task — on knowledge of imposition, the second is connected with optimization of JS code, and the third checks ability to study and apply something new. This time arriving it will be necessary to deal with Audio.API and to write player. Under cat you will find not only analyses of tasks, but also in general useful recommendations both for beginners, and for pilot frontender.
2 years ago
In this report the algorithmic tasks connected from mines pluses by polynoms are considered. More specifically — resolvability of systems of linear mines pluses of polynoms. This task it is polynomial to an equivalent task about determination of the winner in so-called cyclic games (mean payoff games) known to the task lying in intersection of the slozhnostny classes NP and coNP. The second result which is discussed during the report is analog of the theorem of Gilbert of zero for algebra mines pluses.
Mines pluses (or tropical) a half ring is called the set of rational numbers with two operations: mines pluses addition which to eat simply operation of taking of a minimum, and mines pluses with multiplication which is normal addition. Polynoms over mines plus a half ring are defined by analogy with classical polynoms. In essence, mines pluses the polynom sets piecewise linear function from the variables. A root of a polynom is called the point of not smoothness of this function.
The report was read at the faculty of computer sciences opened in Higher School of Economics National Research University with assistance of Yandex. The lecturer Vladimir Podolsky is the senior research associate of Mathematical institute of V.A. Steklov. On FKN gives lectures and conducts seminars within the course "Discrete Mathematics". The report is based on collaborations with Dmitry Grigoriev.
2 years ago
Сегодня завершился финал Яндекс.Алгоритма — ежегодного чемпионата по спортивному программированию, который организует Яндекс. В 2015 году состязание проходило полностью в онлайне — на платформе Яндекс.Контест. Заявки на участие подали программисты из 73 стран. Больше всего участников — из России, Украины, Беларуси, Казахстана, Индии, США, Японии и Китая, но вообще география чемпионата крайне обширна — Бразилия, Индонезия, Перу, Доминиканская Республика, Мозамбик, Сенегал, Каймановы острова. 8,9% зарегистрировавшихся — девушки. Примерно половина всех участников — студенты. Всего мы получили заявки от 3722 человек, из которых до финала дошли 28.
А победителем Яндекс.Алгоритма-2015 стал Геннадий Короткевич. Он по привычке показал лучший результат, решив в финальном раунде пять из шести задач и получив при этом 80 минут штрафного времени. Геннадий занимал первое место в чемпионате Яндекса и в 2013, и в 2014 годах.
Второе место занял Пётр Митричев, а третье — Евгений Капун. Они решили по четыре задачи, при этом Пётр набрал 31 штрафную минуту, а Евгений — 79 минут. Результаты всех финалистов можно посмотреть на сайте Яндекс.Алгоритма.
Задачи для Яндекс.Алгоритма составляет международная команда, в которую входят как сотрудники Яндекса, так и приглашённые эксперты — в том числе победители и финалисты состязаний ACM ICPC и Topcoder Open. И мы по традиции подготовили для вас разборы всех заданий. Решить все из них никому не удалось. Больше всего участников справились с задачей B, а вот задания A и D решило всего по одному человеку.
Sooner or later each of us reflects, whether it makes sense to pass on to children the experience. Today it is possible to tell that the project which we have begun more than two years ago, has taken place and looks successful. Time result positive, we have solved why would not will share with dear public.
On August 1, 2015 in the city of Omsk the Russia's first tournament on leanpoker as which organizer Ivan Nemytchenko has acted has taken place. Tournament on leanpoker has taken place in IT Loft to the address Omsk, Uchebnaya St. 83. Leanpoker — it not simple competition of players in Unlimited Texas Holdem. This competition of programmers who had to write poker bot which purpose was to change bots rivals. Feature of tournament also is also that there is no basic difference in in what programming language the team writes. Therefore and programming languages in tournament it has been provided a little, it is slightly less, than number of teams.
2 years ago
In practice often there is problem of distribution of objects or people in couples with each other. For example, distribution of employees on vacancies, forming of committees, distribution of entrants on higher education institutions. Today's lecture is devoted to the theory and practice of creation of mechanisms of such distribution taking into account preferences of individuals. It has been read at the faculty of computer sciences opened in the Tower with assistance of Yandex.
Lecturer? Sofya Gennadyevna Kiselgof, junior researcher of the International scientific laboratory of the analysis and choice of solutions of NIU HSE. Teacher of department of mathematics of economics department. At faculty of computer sciences gives the course Operations Research and Game Theory. Has defended the master's thesis on subject? The generalized matchings at the preferences which are not linear orders?. Sofya Gennadyevna conducted research of the mechanism of transfer of entrants in the Russian higher education institutions as a result of which the model describing behavior of the entrant at choice of higher education institution has been constructed.
Last year I have decided to be engaged in such very widespread activity as data record of the activity. I have noted that over the past few years my trips became quicker and were given me more simply as the season came nearer to the end so it became interesting to me to make sure available similar improvements of the physical form. Using only the smartphone and the corresponding applications, I have written 27 trips between the house and work, and then used Wolfram Language for reading, the analysis and visualization of results.
Here the image about Google Earth showing my morning bicycle route having distance is slightly less than 18 km, lying from the East on the West.