Although reinforcement learning has been extensively modeled, few agent models that incorporate values use biologically plausible neural networks as a uniform computational architecture. We call biologically plausible neural network architecture neuromorphic. This paper discusses some theoretical constraints on neuromorphic intrinsic value systems [3]. By intrinsic, we mean a value system that is likely programmed by the genes, whose value bias has already taken a shape at the birth time. Such an intrinsic value system plays an important role in developing extrinsic values through the agent's own experience during its life span. Based on our theoretical constraints, we model two types of neurotransmitters, serotonin and dopamine, to construct a neuromorpic intrinsic value system based on a uniform neural network architecture. Serotonin represents punishment and stress, while dopamine represents reward and pleasure. Experimentally, this model allows our simulated robot to develop an attachment to one entity and fear another.

This paper describes the design of an archival temperature monitoring tag to be used to collect habitat information of Atlantic salmon during the marine phase of the species' natural history. The monolithic circuit should not exceed an average power dissipation of 3.4 /spl mu/W, so that a small 3V, 30 mAh lithium battery will provide the device with the expected lifetime of 3 years. The thermal sensor is realized by a p-n-junction, The recorded junction voltages are digitized and stored in static RAM cells. The tag is programmed before deployment by an external computer via a serial interface. The device's temperature-monitoring protocol is defined at this stage. Upon retrieval of the tag, the stored temperature values will be transferred to a computer where the geographical location of the fish at the actual sampling time will be deduced through retrospective navigation based on known sea temperature distribution.

This paper describes the design of an archival temperature monitoring tag to be used to collect habitat information of Atlantic salmon during the marine phase of the species' natural history. The monolithic circuit should not exceed an average power dissipation of 5 /spl mu/W so that a small 3 V lithium battery of 50 mAh charge will provide the device with the expected lifetime of 3 years. The thermal sensor is realized by a pn-junction. The recorded junction voltages are digitized and stored in static RAM cells. Upon retrieval of the tag, the stored temperature values will be transferred to a computer where the geographical location of the fish at the actual sampling time will be deduced through retrospective navigation based on known sea temperature distribution.

We present a strongly code disjoint (SCD) built-in current sensor (BICS) based on self-exercising concept. The integration of this SCD BICS with a self-checking circuit achieves the totally self-checking goal in static CMOS realizations, even in the presence of stuck-on and bridging faults, and results in a strongly fault-secure realization. Low-cost and high fault coverage is attractive for many high reliability and critical applications.>

CMOS is a popular technology today for very large scale integrated (VLSI) circuits. But, conventional functional testing cannot guarantee the detection of some defects. Built-in current testing has been suggested to enhance the defect coverage. In this paper, the authors present a high-speed built-in current sensing (BICS) circuit design. An experimental CMOS VLSI chip containing BICS is described. The power bus current of an 8*8 parallel multiplier is monitored. This BICS detects all implanted short circuit defects and some open circuit defects at a clock speed of 30 MHz (limited by the test set up). SPICE3 simulations indicate a defect detection time of 2 ns.>

In analog frequency-division multiplexed fiber optic systems, laser nonlinearities transfer energy between the different carriers, resulting in intermodulation distortion and interference. A third-order polynomial without memory is used to model the nonlinearity. The FM frequency-division multiplexed (FM/FDM) signal consists of a set of FM carriers at frequencies.N_{\omega 0}, whereN = L, L + 1, L +2, ... , M. The video distortion introduced by the nonlinearity is determined for various types of second- and third-order intermodulation terms. The intermodulation distortion frequency spectrum is determined for FM signals with assumed Gaussian spectra.

Ray position in slope-controlled waveguides is studied. A linear combination of three consecutive position measurements is used to calculate the slope error of the middle lens. This is the error input to the middle lens positioning system. Independent lens correction results and thus a decoupling of space and time response is achieved. The system is shown to be stable. Overshoot in the transient response due to cross coupling between controllers is eliminated. Performance is compared to that of the position-controlled waveguide. Computer simulations indicate that the system is stable in the presence of statistical parameter variations.

Beam position control in beam waveguides subject to statistical variations is studied. Variations in alignment and component values cause the beam to be displaced from the guide axis. A control system to correct this displacement is described. Both the expected and the rms values of the beam position are discussed. An approximate expression for the stability conditions is determined.