Adaptive multi-teacher multi-level knowledge distillation
نویسندگان
چکیده
منابع مشابه
Sequence-Level Knowledge Distillation
Neural machine translation (NMT) offers a novel alternative formulation of translation that is potentially simpler than statistical approaches. However to reach competitive performance, NMT models need to be exceedingly large. In this paper we consider applying knowledge distillation approaches (Bucila et al., 2006; Hinton et al., 2015) that have proven successful for reducing the size of neura...
متن کاملFuzzy Adaptive Granulation Multi-Objective Multi-microgrid Energy Management
This paper develops an energy management approach for a multi-microgrid (MMG) taking into account multiple objectives involving plug-in electric vehicle (PEV), photovoltaic (PV) power, and a distribution static compensator (DSTATCOM) to improve power provision sharing. In the proposed approach, there is a pool of fuzzy microgrids granules that they compete with each other to prolong their lives...
متن کاملAn Adaptive Multi-level Information Filtering System
To conduct efficient information filtering, uncertainties occurring at multiple levels must be managed. Uncertainties can occur due to changing document space as well as stochasticity and non-stationarity of the user. A new filtering model based on established techniques in information retrieval and artificial intelligence is proposed. These techniques include document representation by vector-...
متن کاملAdaptive multi-level explicit congestion notification
In this paper we extend Multi-level ECN, a new TCP congestion scheme, which we have proposed previously. The Multi-level Explicit Congestion Notification (MECN) algorithm allows network operators to achieve high throughput with corresponding low delays. But MECN average queue, is sensitive to its parameter settings and its the level of congestion, hence no guarantees can be given about delay. D...
متن کاملMULAN: Multi-Level Adaptive Network Filter
A security engine should detect network traffic attacks at line-speed. When an attack is detected, a good security engine should screen away the offending packets and continue to forward all other traffic. Anomaly detection engines must protect the network from new and unknown threats before the vulnerability is discovered and an attack is launched. Thus, the engine should integrate intelligent...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Neurocomputing
سال: 2020
ISSN: 0925-2312
DOI: 10.1016/j.neucom.2020.07.048