Thuật ngữ AI
Từ điển đầy đủ về Trí tuệ nhân tạo
Analog VLSI
Very Large Scale Integration circuits using continuous rather than discrete signals to process information. These circuits enable a more natural implementation of biological neuronal behaviors with significantly reduced energy consumption.
Subthreshold Operation
Operation of MOS transistors below their threshold voltage to exploit the exponential current-voltage relationship. This technique allows ultra-low-power analog circuits that simulate the characteristics of biological neurons.
Current-Mode Computing
Computing paradigm where information is encoded and processed in the form of electrical currents rather than voltages. This approach enables natural arithmetic operations like addition through simple current summation.
Operational Transconductance Amplifier
Amplifier whose output is a current proportional to the input voltage, characterized by its transconductance parameter gm. A fundamental element in neuromorphic circuits for implementing synaptic weighting and integration functions.
Differential Pair
Configuration of two transistors whose output currents vary in opposition based on the differential input voltage. Used in neuromorphic circuits to implement sigmoid functions and analog comparators.
Floating Gate
Isolated electrode in a MOS transistor capable of storing electrical charges for long periods. Used in analog neuromorphic circuits as a non-volatile memory element for synaptic weights.
Continuous-Time Computing
Computing approach where operations are performed in real time without temporal discretization, unlike clock-based digital systems. Enables asynchronous processing closer to natural brain functioning.
Current Mirror
Circuit that copies an input current to one or more outputs with a precise replication relationship. Fundamental in neuromorphic circuits for bias current distribution and synaptic weight implementation.
Analog Memory Cell
Memory cell capable of storing and retrieving continuous values rather than discrete binary states. Essential in neuromorphic systems to represent synaptic weights with fine granularity.
Charge Pump
Circuit that transfers electrical charges between nodes to generate reference voltages or currents. Used in neuromorphic circuits to implement learning mechanisms based on charge modulation.
Resistive Switching
Phenomenon by which the resistance of a material can be reversibly modified by applying an electrical voltage. Operating basis of resistive memories (RRAM) used as programmable artificial synapses.