Galaxy morphological classificationGalaxy morphological classification is a system used by astronomers to divide galaxies into groups based on their visual appearance. There are several schemes in use by which galaxies can be classified according to their morphologies, the most famous being the Hubble sequence, devised by Edwin Hubble and later expanded by Gérard de Vaucouleurs and Allan Sandage. However, galaxy classification and morphology are now largely done using computational methods and physical morphology.
ProtogalaxyIn physical cosmology, a protogalaxy, which could also be called a "primeval galaxy", is a cloud of gas which is forming into a galaxy. It is believed that the rate of star formation during this period of galactic evolution will determine whether a galaxy is a spiral or elliptical galaxy; a slower star formation tends to produce a spiral galaxy. The smaller clumps of gas in a protogalaxy form into stars. The term "protogalaxy" itself is generally accepted to mean "Progenitors of the present day (normal) galaxies, in the early stages of formation.
Transformer (machine learning model)A transformer is a deep learning architecture that relies on the parallel multi-head attention mechanism. The modern transformer was proposed in the 2017 paper titled 'Attention Is All You Need' by Ashish Vaswani et al., Google Brain team. It is notable for requiring less training time than previous recurrent neural architectures, such as long short-term memory (LSTM), and its later variation has been prevalently adopted for training large language models on large (language) datasets, such as the Wikipedia corpus and Common Crawl, by virtue of the parallelized processing of input sequence.
Starburst galaxyA starburst galaxy is one undergoing an exceptionally high rate of star formation, as compared to the long-term average rate of star formation in the galaxy or the star formation rate observed in most other galaxies. For example, the star formation rate of the Milky Way galaxy is approximately 3 M☉/yr, while starburst galaxies can experience star formation rates of 100 M☉/yr or more. In a starburst galaxy, the rate of star formation is so large that the galaxy will consume all of its gas reservoir, from which the stars are forming, on a timescale much shorter than the age of the galaxy.
Expert systemIn artificial intelligence, an expert system is a computer system emulating the decision-making ability of a human expert. Expert systems are designed to solve complex problems by reasoning through bodies of knowledge, represented mainly as if–then rules rather than through conventional procedural code. The first expert systems were created in the 1970s and then proliferated in the 1980s. Expert systems were among the first truly successful forms of artificial intelligence (AI) software.
Galaxy ZooGalaxy Zoo is a crowdsourced astronomy project which invites people to assist in the morphological classification of large numbers of galaxies. It is an example of citizen science as it enlists the help of members of the public to help in scientific research. There have been 15 versions as of July 2017. Galaxy Zoo is part of the Zooniverse, a group of citizen science projects. An outcome of the project is to better determine the different aspects of objects and to separate them into classifications.
First observation of gravitational wavesThe first direct observation of gravitational waves was made on 14 September 2015 and was announced by the LIGO and Virgo collaborations on 11 February 2016. Previously, gravitational waves had been inferred only indirectly, via their effect on the timing of pulsars in binary star systems. The waveform, detected by both LIGO observatories, matched the predictions of general relativity for a gravitational wave emanating from the inward spiral and merger of a pair of black holes of around 36 and 29 solar masses and the subsequent "ringdown" of the single resulting black hole.
Deep learningDeep learning is part of a broader family of machine learning methods, which is based on artificial neural networks with representation learning. The adjective "deep" in deep learning refers to the use of multiple layers in the network. Methods used can be either supervised, semi-supervised or unsupervised.
Gravitational-wave observatoryA gravitational-wave detector (used in a gravitational-wave observatory) is any device designed to measure tiny distortions of spacetime called gravitational waves. Since the 1960s, various kinds of gravitational-wave detectors have been built and constantly improved. The present-day generation of laser interferometers has reached the necessary sensitivity to detect gravitational waves from astronomical sources, thus forming the primary tool of gravitational-wave astronomy.
Active learning (machine learning)Active learning is a special case of machine learning in which a learning algorithm can interactively query a user (or some other information source) to label new data points with the desired outputs. In statistics literature, it is sometimes also called optimal experimental design. The information source is also called teacher or oracle. There are situations in which unlabeled data is abundant but manual labeling is expensive. In such a scenario, learning algorithms can actively query the user/teacher for labels.
Residual neural networkA Residual Neural Network (a.k.a. Residual Network, ResNet) is a deep learning model in which the weight layers learn residual functions with reference to the layer inputs. A Residual Network is a network with skip connections that perform identity mappings, merged with the layer outputs by addition. It behaves like a Highway Network whose gates are opened through strongly positive bias weights. This enables deep learning models with tens or hundreds of layers to train easily and approach better accuracy when going deeper.
Supervised learningSupervised learning (SL) is a paradigm in machine learning where input objects (for example, a vector of predictor variables) and a desired output value (also known as human-labeled supervisory signal) train a model. The training data is processed, building a function that maps new data on expected output values. An optimal scenario will allow for the algorithm to correctly determine output values for unseen instances. This requires the learning algorithm to generalize from the training data to unseen situations in a "reasonable" way (see inductive bias).
