Memristive synapses for spike timing- and rate-based synaptic plasticity

Memristors, due to their special features, i.e. non-volatility and the ability to be programmed while operating, have significantly attracted neuromorphic engineers attention in the course of last five years. They have been extensively utilised to implement well-known synaptic plasticity learning algorithms such as pair-based STDP and SRDP [1-3]. Attempts are also made to mimic experimental outcomes of higher order spike-based synaptic plasticity rules such as the suppressive STDP rule of Froemke et al. [2,4]. However, the realization of rate-based STDP experiments [5] using a memristive synapse is rather unexplored. Although a VLSI synapse which can account for a range of synaptic plasticity experiments including Pair-based, Triplet, Qudruplet, rate-based STDP, and Bienenstock Booper Munro (BCM) has already been proposed [6], a memristive synapse that has similar plasticity capabilities is yet to be developed.

In this workgroup we will discuss and review previous implementations of memristive synapses. We will also talk about the applications and challenges in the design of neuromorphic memristive systems. As a project, we will collaborate on the design of a new memristive synapse, which aims at improving the synaptic plasticity (learning) capability of previously proposed devices. In particular, we will work toward implementing a memristive synapse, which can replicate a number of interesting features of biological synapses, including the spike-rate based plasticity. In addition, we test the synapse against implementing higher order STDP rules, such as triplet-based STDP.

For our project, we are going to use Jean-Pascal Pfister's triplet STDP model presented in [7]. Interestingly, Jean-Pascal is also contributing to our workgroup! So this is a good opportunity to get something done. Even if we cannot complete what we started during the workshop, I intend to continue working on this idea, hopefully with your help and collaborations.



Possible activities

  • Developing a memristive synapse to account for triplet of spikes rather than spike pairs
  • Testing the proposed synapse for replicating a number of timing-based experimental data, such as those of Triplet and Quadruplet experiments
  • Testing the proposed synapse for replicating a number of rate-based experimental data, such as Frequency-dependent pairing experiment in the visual cortex, as well as the BCM
  • Testing the proposed synapse in a simple classification task
  • Any other ideas are highly appreciated!


  • [1] Spike-Based Synaptic Plasticity in Silicon: Design, Implementation, Application, and Challenges
  • [2] Plasticity in memristive devices for spiking neural networks
  • [3] On spike-timing-dependent-plasticity, memristive devices, and building a self-learning visual cortex
  • [4] Neuronal Synapse as a Memristor: Modeling Pair- and Triplet-Based STDP Rule
  • [5] Rate, Timing, and Cooperativity Jointly Determine Cortical Synaptic Plasticity
  • [6] A Neuromorphic VLSI Design for Spike Timing- and Rate-based Synaptic Plasticity
  • [7] Triplets of Spikes in a Model of Spike Timing-Dependent Plasticity


For simulating the memristive synapse model, one can use Matlab or any other programming language.


Achieved results using the newly developed memristive circuit for the triplet STDP can reproduce the frequency-dependent pairing experiments, where the available pair-based STDP fails.

See the following two figures:

In the first figure, the frequency-dependent pairing experiment has been performed using the available pair-based memristive STDP circuit and the results are shown in red. The figure depicts the failure of the pair-based memristive circuit to mimic the results obtained in the experiments for two different \Delta t.

The second figure, on the other hand, demonstrates the ability of the proposed triplet STDP memristive circuit developed in this workgroup, in replicating a close match to the targeted experimental data.

Other ideas and people involved for further collaboration beyond the workshop:

  • Looking at other memristor models to implement the triplet rule, e.g. 2nd order memristors as discussed in our workgroup (Selina, Manuel, Abu and Fabien)
  • Implementing rate-based BCM learning model using this timing-based model
  • Utilizing this model for a cognitive task, such as pattern classification (Rawan, Chris, Damir)
  • Utilizing the proposed model in an X-bar structure and building up a network
Last modified 4 years ago Last modified on 05/14/15 15:19:43

Attachments (2)

Download all attachments as: .zip