Dictionary
Concept dictionary
The vocabulary you need to read this archive. Each entry is the term as the field uses it, paired with a paper in the corpus where Hazan works with it.
The definitions are ours, written to get you into the papers — they are not quotations of Hazan. The citation under each one is the part that is his.
- Reservoir computing
- A family of recurrent-network methods in which a large, fixed, randomly connected dynamical system — the reservoir — projects an input into a high-dimensional state, and only a simple readout is trained. The reservoir is not learned; its dynamics are the computation.
- In the corpus Stability and Topology in Reservoir Computing (2010) ↗
- Liquid state machine (LSM)
- The spiking-neuron form of reservoir computing: a recurrent pool of spiking neurons whose transient activity — the "liquid" — encodes the recent history of an input stream, read out by a trained classifier. Introduced by Maass and colleagues; Hazan works on them, and has since 2010.
- In the corpus Topological constraints and robustness in liquid state machines (2011) ↗
- Topological constraint
- A restriction on how the units inside a reservoir may connect to one another — for instance, wiring that respects locality on a lattice rather than connecting at random. A recurring result in this archive is that such constraints bear on how robust the reservoir is to damage in its own components.
- In the corpus Topological constraints and robustness in liquid state machines (2011) ↗
- Spiking neural network (SNN)
- A neural network whose units communicate with discrete, time-stamped events (spikes) rather than continuous activation values, so that timing itself carries information. The dominant substrate across this corpus.
- In the corpus Locally connected spiking neural networks for unsupervised feature learning (2019) ↗
- BindsNET
- The open-source spiking-neural-network simulation project Hazan led — described on his Allen Discovery Center profile as "a state-of-the-art framework designed for rapid constructions of rich simulations of spiking networks." The 0.3.4 release deposit in this corpus lists him first among its contributors.
- In the corpus BindsNET/bindsnet: 0.3.4 (software release record) ↗
- Dynamic clamp
- An electrophysiology technique in which a computer reads a living neuron’s membrane potential and injects a current calculated in real time, letting an experimenter add a simulated ion conductance to a real cell. Used here to map the excitability phase diagram of the Hodgkin–Huxley model.
- In the corpus Dynamic clamp constructed phase diagram for the Hodgkin and Huxley model of excitability (2020) ↗
- Closed-loop experiment
- An experimental setup in which stimulation delivered to living tissue depends on what that tissue just did — recording and stimulation joined in a real-time loop rather than run open-loop. Hazan built such a platform for cortical neuronal networks at the Technion.
- In the corpus Closed Loop Experiment Manager (CLEM)—An Open and Inexpensive Solution for Multichannel Electrophysiological Recordings and Closed Loop Experiments (2017) ↗
- Bioelectric circuit
- A network of cells coupled by voltage state and ion-channel dynamics, treated as a computational circuit that carries information above the level of the single cell. The seam between this archive and Michael Levin’s work on collective and basal intelligence.
- In the corpus Exploring the Behavior of Bioelectric Circuits Using Evolution Heuristic Search (2022) ↗
- Evolutionary heuristic search
- Optimisation by maintaining a population of candidate solutions and iteratively selecting, recombining, and perturbing them — used in this archive both to train spiking networks and to probe the behaviour space of bioelectric circuits.
- In the corpus Training spiking neuronal networks to perform motor control using reinforcement and evolutionary learning (2022) ↗
- Memristor
- A two-terminal device whose resistance depends on the history of charge that has passed through it, making it a candidate physical substrate for synaptic weight. Applied here to building a liquid state machine in hardware with in-situ training.
- In the corpus Memristor Based Liquid State Machine With Method for In-Situ Training (2024) ↗