Unit:
Κατεύθυνση Μεγάλα Δεδομένα και Τεχνητή ΝοημοσύνηΠληροφορική
Author:
Tsapelas Christos
Supervisors info:
Γεωργία Κούτρικα, Διευθύντρια Έρευνας, Κέντρο Έρευνας και Καινοτομίας, ”ΑΘΗΝΑ”
Γιάννης Ιωαννίδης, Καθηγητής, Τμήμα Πληροφορικής και Τηλεπικοινωνιών, Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών
Τιμολέων Σελλής , Διευθυντής Έρευνας, Μονάδα ”ΑΡΧΙΜΗΔΗΣ”, Κέντρο Έρευνας στην Τεχνητή Νοημοσύνη, την Επιστήμη Δεδομένων και τους Αλγορίθμους , Ερευνητικό Κέντρο Αθηνά
Original Title:
QPSeeker - An efficient Neural Planner combining both data and queries through Variational Inference
Translated title:
QPSeeker - An efficient Neural Planner combining both data and queries through Variational Inference
Summary:
Query optimization is a well-studied problem in the database community. Recently, deep learning methods have been applied either to assist the query optimizer on measures like cardinality estimation, computational cost prediction and query execution time predic- tion or implementing neural-based optimizers from scratch. Despite the promising results, very few tackle multiple aspects of the optimizer at the same time or combine both the underlying data and a query workload. QPSeeker takes a step towards a neural database planner, encoding first the information provided from the workload and second the under- lying tables, using the power of special-designed language models for tabular data. Next, it applies a special form of attention to combine these two sources to approximate the distributions of cardinalities, costs and execution times of possible query plans. At infer- ence, when a query is submitted to the database, QPSeeker uses its learned cost model and traverses the query plan space using Monte Carlo Tree Search to provide the best execution plan for the query.
Main subject category:
Technology - Computer science
Keywords:
machine learning, multi-modal cross-attention, variational inference, cardinality estimation, cost estimation, query latency prediction
