Showing 1–2 of 2 results for author: DePasquale, B
-
full-FORCE: A Target-Based Method for Training Recurrent Networks
Authors:
Brian DePasquale,
Christopher J. Cueva,
Kanaka Rajan,
G. Sean Escola,
L. F. Abbott
Abstract:
Trained recurrent networks are powerful tools for modeling dynamic neural computations. We present a target-based method for modifying the full connectivity matrix of a recurrent network to train it to perform tasks involving temporally complex input/output transformations. The method introduces a second network during training to provide suitable "target" dynamics useful for performing the task.…
▽ More
Trained recurrent networks are powerful tools for modeling dynamic neural computations. We present a target-based method for modifying the full connectivity matrix of a recurrent network to train it to perform tasks involving temporally complex input/output transformations. The method introduces a second network during training to provide suitable "target" dynamics useful for performing the task. Because it exploits the full recurrent connectivity, the method produces networks that perform tasks with fewer neurons and greater noise robustness than traditional least-squares (FORCE) approaches. In addition, we show how introducing additional input signals into the target-generating network, which act as task hints, greatly extends the range of tasks that can be learned and provides control over the complexity and nature of the dynamics of the trained, task-performing network.
△ Less
Submitted 9 October, 2017;
originally announced October 2017.
-
Using Firing-Rate Dynamics to Train Recurrent Networks of Spiking Model Neurons
Authors:
Brian DePasquale,
Mark M. Churchland,
L. F. Abbott
Abstract:
Recurrent neural networks are powerful tools for understanding and modeling computation and representation by populations of neurons. Continuous-variable or "rate" model networks have been analyzed and applied extensively for these purposes. However, neurons fire action potentials, and the discrete nature of spiking is an important feature of neural circuit dynamics. Despite significant advances,…
▽ More
Recurrent neural networks are powerful tools for understanding and modeling computation and representation by populations of neurons. Continuous-variable or "rate" model networks have been analyzed and applied extensively for these purposes. However, neurons fire action potentials, and the discrete nature of spiking is an important feature of neural circuit dynamics. Despite significant advances, training recurrently connected spiking neural networks remains a challenge. We present a procedure for training recurrently connected spiking networks to generate dynamical patterns autonomously, to produce complex temporal outputs based on integrating network input, and to model physiological data. Our procedure makes use of a continuous-variable network to identify targets for training the inputs to the spiking model neurons. Surprisingly, we are able to construct spiking networks that duplicate tasks performed by continuous-variable networks with only a relatively minor expansion in the number of neurons. Our approach provides a novel view of the significance and appropriate use of "firing rate" models, and it is a useful approach for building model spiking networks that can be used to address important questions about representation and computation in neural systems.
△ Less
Submitted 27 January, 2016;
originally announced January 2016.
