Winner-Take-All Networks


  • The winner--take--all (WTA) circuit models a neural network consisting of n excitatory cells and one inhibitory cell. When the excitatory cells are active, they excite the inhibitory cell which in turn inhibits the excitatory cells. The inhibitory cell's activity will increase until it kills off all the excitatory cells except one. If the loop gain is high enough, this excitatory cell will be able to maintain the requisite level of inhibitory cell activity by itself. Naturally, the excitatory cell that survives will be the one with the largest extrinsic input.
  • A useful extension of the WTA network is to introduce a pool of inhibitory neurons with local connectivity between them. Then competition occurs over a local region whose extent is determined by how far signals spread among the inhibitory cells. Excitatory cells that are outside this region will not be inhibited much. In these cases, we talk about the locality of these WTA networks.

Reading / Resources:

  • Original Paper on Winner-Take-All Circuits as well as the longer technical Report
  • A more than necessary detailed overview on scanners
  • For those who want to see an application of this circuit, which we will discuss more in a later section, I would recommend reading this paper
  • For those of you who have not had enough floating-gate circuits yet, we have another paper which extends the Winner-Take-All concept.
  • Lecture Notes for Lecture 12: 1 , 2 , 3 , 4 , Lecture 13: 1 , 2 , 3 , Lecture 14: 1 , 2 , 3 , 4 ,

Project Questions:

In this project, we will focus on one type of classifier structure, the Winner-Take-All (WTA) concept, that derives its function from neurobiological models. We will find the WTA compiles well into the SoC FPAA structure.

TIA Characterization

The TIA is a fundamental block for an FPAA based system, particularly when not wanting to utilize an ammeter (i.e. static) for every measurement. The purpose of the TIA is to use the on-chip OTAs and fgOTAs to create a linear current-to-voltage converter. You will want to program the bias current of the fgOTAs to be meaningful for the currents that you are measuring. The basic block diagram looks like

where the bias current of the OTA (non FG) would be fine to put at a few uA (like 1-4uA). The feedback element will be a FG OTA device. The bias current you program will set the effective current-to-voltage gain for your measurement. You might notice you explicitly have this block in the library.

Characterizing the TIA experiment : For these measurements, we first apply an input current by attaching a large resistor (i.e. 10MOhm or larger) to the input terminal of the TIA. Yes, this would require a board. Sweep the resistor's other terminal by changing the voltage of a DAC, and measure the resulting output voltage.

Plot the output voltage as a function of the input current you are applying, using circles to designate that this is measured data. Calculate the theoretical transconductance of the fgOTA in feedback, and plot the theoretical output voltage as a function of the input current. Use a line to indicate this is theoretical data. So for example, your plot command might look like plot(iin,vout,`o',iin,vout1).

Transistor characterization of the TIA : Hook up your TIA to a MOSFET on-chip, ideally the golden FET (the device already characterized) such that the TIA is connected to the drain of that device, with the source at the power supply (GNd for an nFET, Vdd for a pFET). Sweep the gate voltage for several relevant points and look at the TIA output. You should see a clear part of the I-V curve, as well as what happens when the current is too small or too large given the bias current you would use for the feedback OTA device.

Two-input Winner-Take-All (WTA) circuit.

This set of measurements you want to compile a basic two input WTA block (Fig. 2). Use two pFET devices to input the currents into this circuit, and use two pFET devices that are diode connected to measure the effective current going between both legs of this device. The figure below shows the resulting circuit:

Choose a current range (e.g. 10nA) for your measurements. Do a sweep of currents around that range, and look at the output current change with input current change. Remember, you are setting voltages and measuring voltages and will need to convert between representations.

In this experiment, we will measure the gain of the WTA circuit. The maximum gain is very high---a five percent difference in current is enough to win. This is equivalent to a 1.5mV difference in gate voltages. Perform a tight sweep near where the input currents cross to see the resulting gain region as closely as you can measure. You have done multiple measurements at this point, and probably have a few possible techniques for this measurement. Try to determine the resulting current gain of this structure from measured data, and compare to what you expect to experimentally measure. Finally, discuss the effects of mismatch (threshold voltage mismatch) in this circuit, particularly in terms of any observed offsets.

Many-input Winner-Take-All (WTA) circuit.

Next, we want to look at a larger WTA system between 5 and 7 inputs.

  • Use pFET devices (FG pFETs are good) to transform the output currents to a voltage through a comparison operation. The WTA block will have high-gain digital outputs (or near digital outputs). If possiblem, set up your WTA block for 1 or 2 winning nodes.
  • You will want to use the shift register to look at (scan) the resulting outputs.
  • You should set up your input pattern such that you have two inputs that you can change; you might tie the other inputs to a fixed potential.
You will want to show a couple of plots showing different winning outputs. A case where you can make one input win, a second case where you can make another input win, a case where neither are winning, and a case where the two inputs are close so maybe two devices are near to winning.