Filtern
Dokumenttyp
Sprache
- Englisch (5)
Volltext vorhanden
- ja (5)
Schlagworte
- Automatische Spracherkennung (1)
- Bildung (1)
- Digitalisierung (1)
- Genetischer Algorithmus (1)
- Künstliche Intelligenz (1)
- Maschinelles Lernen (1)
- Medienkompetenz (1)
- Raspberry Pi Zero W (1)
- Shakespeare, William (1)
- Turing-Test (1)
Institut
Digital education: ”education-with” / ”education-about” distinction and the teleological definition
(2020)
Despite being an object of multi-billion public policies and private sector initiatives, the term ”digital education” seems to lack a clear, unambigous, lexicon-ready definition. A closer analysis reveals that in common parlance, the term is used to denote phenomena related to overlapping but distinct topics like ”technology”, ”media” or ”informatics”. Such polysemy implies an overall lack of
clarity which a public debate about education policies should rather avoid. For this reason, we propose to start sorting things out by defining the term ”digital education” in terms of dichotomy of two subordinated concepts, which we label as ”education-about-digital” and ”education-with-digital”. Postulation of this dichotomy combined with analysis of "school's mission" as defined in the legal codices of Land Berlin naturally leads to ”teleological definition” which delimits the concept of digital education in terms of its ideal human result.
Repetition of morphological or lexical units is an established technique able to reinforce the impact of one's argument upon the audience. Rhetoric tradition has canonized dozens of repetition-involving schemas as figures of speech. Our article shows a way how hitherto ignored repetition-involving schemata can be identified. It shows that certain classes of repetitive figures can be represented in terms of specific sequences of integer numbers and vice versa, how specific sets of integer numbers can be translated into sets of regexes able to match repetition-involving expressions. A "Shakespeare number" S is simply defined as an integer with at least one repeated digit in which no digit bigger than X can occur if ever a digit X had not yet occurred in S's decimal representation. Hence, 121 is a Shakespeare number, while 123 or 211 are not. A set of "entangled numbers" is subsequently defined as a subset of "Shakespeare numbers" with an additional property that all digits which occur in them are repeated at least twice in the decimal representation of the number. Thus, a 1212 is an entangled number while 1211 is not. A complete set E of entangled numbers of maximal length of 10 digits is subsequently generated and every member of E is translated into a regex. Each regex is subsequently exposed to all utterances in all works of William Shakespeare, allowing us to pinpoint 3367 instances of 172 distinct E-schemata. This nomenclature may allow scholars to lead a discussion about schemata which have escaped the attention of classical interpretators.
The paper presents a novel method of multiclass classification. The method combines the notions of dimensionality reduction and binarization with notions of category prototype and evolutionary optimization. It introduces a supervised machine learning algorithm which first projects documents of the training corpus into low-dimensional binary space and subsequently uses canonical genetic algorithm in order to find a constellation of prototypes with highest classificatory pertinence. Fitness function is based on a cognitively plausible notion that a good prototype of a category C should be as close as possible to members of C and as far as possible to members associated to other categories. In case of classification of documents contained in a 20-newsgroup corpus into 20 classes, our algorithm seems to yield better results than a comparable deep learning "semantic hashing" method which also projects the semantic data into 128-dimensional binary (i.e. 16-byte) vector space.
The original TuringTest is modified in order to take
into account the age&gender of a Judge who evaluates the
machine and the age&gender of a Human with whom the
machine is compared during evaluation. This yields a basic taxonomy of TuringTest-consistent scenarios which is
subsequently extended by taking into account the type of intelligence being evaluated. Consistently with the Theory of Multiple Intelligences, nine basic intelligence types are
proposed, and an example of a possible scenario for evaluation of emotional intelligence in early stages of development is given.
It is suggested that specific intelligence types can be
subsequently grouped into hierarchy at the top of which is seated an Artificial Intelligence labelled as “meta-modular”. Finally, it
is proposed that such a meta-modular AI should be defined as an Artificial Autonomous Agent and should be given all the rights and responsibilities according to age of human counterparts in
comparison with whom an AI under question has passed the
TuringTest
Abstract. This study compares performance of different speech com-
mand classification systems which can be executed on an Raspberry
Pi Zero ARMv6 architecture. Three systems are evaluated: first one,
TREELITE_MFCC, is based on Treelite system cross-compiles a MFCC-
classifying Random Forest into a highly optimized shared library; second,
TFLITE_MFCC, performs classification of MFCC inputs by means of
convolutional neural network encoded as a lightweight TensorFlow Lite
model while the third one - labeled as TFLITE_RAW - uses more com-
plex network to directly classify the audio signal. We evaluate models
not only in terms of their accuracy and precision, but also in terms of
execution time, voltage, current and energy consumed. We observe that
while TFLITE_RAW offers superior performance in terms of accuracy,
TREELITE_RAW is also worth of consideration for low-latency real-
life applications since it offers relatively good performance (avg. micro-
precision=0.917) but its predictions, when executed on Raspberry Pi
Zero, are significantly faster and cost less energy than TensorFlow Lite
models.