Mining Message Flows using Recurrent Neural Networks for System-on-Chip Designs

Cao, Yuting; Mukherjee, Parijat; Ketkar, Mahesh; Yang, Jin; Zheng, Hao

doi:10.1109/isqed48828.2020.9137001

Cited by 7 publications

(6 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

Section: Discussionmentioning

confidence: 99%

“…We apply an LSTM-based pattern mining [6] and a rulebased data mining approach FlowMiner on the same synthetic trace set. Work [6] extracts flow specification in the form of sequential patterns. We extract patterns for pattern probability as in [6], but could not produce a complete specification for any of the tests flows of our experiment.…”

Section: Discussionmentioning

confidence: 99%

“…Specification mining problem is prevalent both in hardware and software domains. Therefore many works have addressed this specification mining problem both from hardware and software perspective [5], [6]. We study several tools and techniques for trace mining and find that those tools often fail to find high-quality models.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Mining SoC Message Flows with Attention Model

Ahmed¹,

Bardia²,

Zheng³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

High-quality system-level message flow specifications are necessary for comprehensive validation of system-on-chip (SoC) designs. However, manual development and maintenance of such specifications are a daunting task. We propose a disruptive method that utilizes deep sequence modeling with the attention mechanism to infer accurate flow specifications from SoC communication traces. The proposed method can overcome the inherent complexity of SoC traces induced by the concurrent executions of SoC designs that existing mining tools often find extremely challenging. We conduct experiments on five highly concurrent traces and find that proposed approach outperforms several existing state-of-the-art trace mining tools.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Mining SoC Message Flows with Attention Model

Ahmed¹,

Bardia²,

Zheng³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…An example work presented in [16] that utilizes Baysian Inference to interpret the execution model using LTL formula. [17], [18] are example work in this line of work. We will take [17] for our consideration.…”

Section: Classification Of Trace Minersmentioning

confidence: 99%

“…Work [17] (we refer as YLSTM) applies LSTM (Long Short Term Memory based on RNN) to extract sequential patterns from SoC transaction-level traces. LSTM is capable of capturing "long-term" dependencies and has many applications in natural language processing.…”

Section: G Ylstmmentioning

confidence: 99%

Tools and Algorithms for SoC Communication Traces

Ahmed,

Zheng

2021

Preprint

Self Cite

View full text Add to dashboard Cite

In this paper we study seven well-known trace analysis techniques both from the hardware and software domain, and discuss their performance on communication centric system-on-chip (SoC) traces. SoC traces are usually huge in size and concurrent in nature, therefore mining SoC traces poses additional challenges. We provide a hands on discussion of the selected tools/algorithms in terms of the input, output and analysis method they employ. Hardware traces also varies in nature when observed in different level, this work can help developers/academicians to pick up the right techniques for their works. We take advantage of a synthetic trace generator to find the interestingness of the mined outcomes for each tool as well as we work with a realistic GEM5 setup to find the performance of these tools on more realistic SoC traces. Comprehensive analysis of the tools performance and a benchmark trace dataset are also presented.

show abstract