Physically Accurate Learning-based Performance Prediction of Hardware-accelerated ML Algorithms.
Hadi EsmaeilzadehSoroush GhodratiAndrew B. KahngJoon Kyung KimSean KinzerSayak KunduRohan MahapatraSusmita Dey ManasiSachin S. SapatnekarZhiang WangZiqing ZengPublished in: MLCAD (2022)
Keyphrases
- maximum likelihood
- learning algorithm
- em algorithm
- learning process
- orders of magnitude
- learning systems
- reinforcement learning
- prediction algorithm
- prior knowledge
- computationally efficient
- times faster
- data sets
- knowledge acquisition
- learning problems
- inverse reinforcement learning
- automatically learned
- noise tolerant
- data structure
- computational cost
- theoretical analysis
- high quality
- unsupervised learning
- high accuracy
- worst case
- prediction accuracy
- regression model
- significant improvement
- learning tasks
- social networks