Neuromorphic computing with emerging memory devices

Neuromorphic computing with emerging memory devices

by Daniele Ielmini

Applied Sciences & Engineering AI — Models & Research UYQArtificial intelligence (AI)UYQNNeural networks and fuzzy systems
🎙 Daniele Ielmini 👥 2K 📅 January 4, 2019 ⏱ 50 min 👁 58K 📄 expert opinion 🧭 2026-08-18
Available in: English (current) Français

Keywords

neuromorphicRRAMSTDPHopfield networkin-memory computing

Summary

This plenary talk by Prof. Daniele Ielmini at the AI International Conference 2018 presents the motivation and implementation of neuromorphic computing using emerging memory devices. He argues that traditional von Neumann architectures face power and memory walls, necessitating brain-inspired hardware. He introduces resistive switching memories (RRAM) and phase-change memories as synaptic elements, demonstrating spike-timing-dependent plasticity (STDP) in hardware. He shows three toy implementations: a feed-forward perceptron for pattern recognition, a Hopfield network for associative memory (Pavlov’s dog), and a spatiotemporal network for sequence learning. The talk emphasizes the potential of these devices for energy-efficient, in-memory computing, and concludes with future directions.

101 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides valuable insights into the hardware challenges of AI and the potential of emerging memories for neuromorphic computing. The argumentation is solid, grounded in experimental demonstrations and references to published work. The speaker effectively explains the limitations of current architectures and the need for paradigm shift, supporting his claims with concrete examples and results from his research group.

Scientific Rigor, Source Quality, Title Accuracy

The speaker cites several sources, including a Nature Electronics review and IBM’s TrueNorth paper, and mentions industrial efforts by TSMC, Intel, and IBM. The title accurately reflects the content. The talk is a plenary presentation, so it is not peer-reviewed, but the speaker’s expertise and the inclusion of experimental data enhance its credibility.

128 words

Title / Content Match

The title accurately reflects the content, which focuses on neuromorphic computing using emerging memory devices.

Quality & Reliability

8/10

The speaker is a recognized expert in the field, and the talk is based on published research and industrial collaborations. However, the presentation is a plenary talk, not a peer-reviewed publication, and some claims are simplified for a general audience.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • Nature Electronics review on emerging memories — Referenced when discussing types of emerging memories.
  • IBM TrueNorth paper — Referenced when discussing power density and neuromorphic chips.

Concurring Sources

  • Nature Electronics review on emerging memories — Supports the classification of emerging memories.

Contribution & Novelties

The talk provides a comprehensive overview of using emerging memory devices for neuromorphic computing, with concrete hardware implementations. It highlights the potential of RRAM and PCM for synaptic emulation and demonstrates learning algorithms like STDP in hardware.

Pour aller plus loin :

65 words

Radar Profile

The radar profile shows high scores in quantity, quality, and reliability, with a slightly lower technical level, indicating a well-balanced presentation suitable for a broad technical audience.

Reliability 8/10

💬 Sur les 0 commentaires analysés, aucune tendance n'est disponible.