stem/AI/Neural Networks/Properties+Capabilities.md

98 lines
3.2 KiB
Markdown
Raw Normal View History

2023-05-20 01:33:56 +01:00
# Linearity
- Neurons can be linear or non-linear
- Network of non-linear neurons is non-linear
- Non-linearity is distributed
- Helpful if target signals are generated non-linearly
# Input-Output Mapping
- Map input signal to desired response
vault backup: 2023-06-08 17:52:08 Affected files: .obsidian/graph.json .obsidian/workspace-mobile.json .obsidian/workspace.json Food/Meal Plan.md Lab/Linux/Alpine.md Lab/Linux/KDE.md Lab/Scratch Domain.md Lab/Windows/Active Directory.md Languages/Spanish/README.md Languages/Spanish/Spanish.md Money/Accounts.md Money/Monthly/23-04.md Money/Monthly/23-05.md Money/Monthly/23-06.md Projects/Mixonomer.md Projects/NoteCrawler.md Projects/Projects.md Projects/README.md Projects/Selector.md Projects/To Do App.md Projects/img/selector-arch.png STEM/AI/Classification/Supervised/SVM.md STEM/AI/Neural Networks/CNN/FCN/Super-Resolution.md STEM/AI/Neural Networks/CNN/GAN/GAN.md STEM/AI/Neural Networks/CNN/GAN/cGAN.md STEM/AI/Neural Networks/CNN/Interpretation.md STEM/AI/Neural Networks/Deep Learning.md STEM/AI/Neural Networks/MLP/MLP.md STEM/AI/Neural Networks/Properties+Capabilities.md STEM/AI/Neural Networks/RNN/Representation Learning.md STEM/AI/Pattern Matching/Markov/Markov.md STEM/AI/Searching/Informed.md STEM/AI/Searching/README.md STEM/AI/Searching/Searching.md STEM/AI/Searching/Uninformed.md STEM/CS/Languages/Javascript.md STEM/CS/Languages/Python.md STEM/CS/Languages/dotNet.md STEM/CS/Resources.md STEM/IOT/Cyber-Physical Systems.md STEM/IOT/Networking/Networking.md STEM/IOT/Networking/README.md STEM/IOT/Software Services.md STEM/img/cyberphysical-social-data.png STEM/img/cyberphysical-system-types.png STEM/img/cyberphysical-systems.png STEM/img/depth-first-cons.png STEM/img/depth-first.png STEM/img/iot-mesh-network.png STEM/img/iot-network-radar.png STEM/img/iot-network-types 1.png STEM/img/iot-network-types.png STEM/img/markov-start-end-matrix.png STEM/img/markov-start-end-probs.png STEM/img/markov-start-end.png STEM/img/markov-state-duration.png STEM/img/markov-state.png STEM/img/markov-weather.png STEM/img/search-bidirectional.png STEM/img/search-breadth-first.png STEM/img/search-lim-goal.png STEM/img/search-lim1.png STEM/img/search-lim2.png STEM/img/search-lim3-2.png STEM/img/search-lim3.png STEM/img/search-lim4.png STEM/img/searching-graph-tree.png STEM/img/searching-graph.png Work/Freelancing.md
2023-06-08 17:52:09 +01:00
- [Supervised](../Learning.md#Supervised) learning
2023-05-20 01:33:56 +01:00
- Similar to non-parametric statistical inference
- Non-parametric as in no prior assumptions
- No probabilistic model
# Adaptivity
- Synaptic weights
- Can be easily retrained
- Stationary environment
- Essential statistics can be learned
- Model can then be frozen
- Non-stationary environments
2023-05-20 01:33:56 +01:00
- Can change weights in real-time
- In general, more adaptive = more robust
- Not always though
- Short time-constant system may be thrown by short-time spurious disturbances
- Stability-plasticity dilemma
- Not equipped to track statistical variations
- Adaptive system
- Linear adaptive filter
- Linear combiner
- Single neuron operating in linear mode
- Mature applications
- Nonlinear adaptive filters
- Less mature
- Environments typically considered pseudo-stationary
- Speech stationary over short windows
- Retrain network at regular intervals to account for fluctuations
- E.g. stock market
- Train network on short time window
- Add new data and pop old
- Slide window
- Retrain network
2023-05-20 01:33:56 +01:00
# Evidential Response
- Decisions are made with evidence not just declared
- Confidence value
# Contextual Information
- [Knowledge](Neural%20Networks.md#Knowledge) represented by structure and activation weight
2023-05-20 01:33:56 +01:00
- Any neuron can be affected by global activity
- Contextual information handled naturally
# Fault Tolerance
- Hardware implementations
- Performance degrades gracefully with adverse conditions
- If some of it breaks, it won't cause the whole thing to break
- Like a real brain
# VLSI Implementability
- Very large-scale integration
- Chips with millions of transistors (MOS)
- E.g. microprocessor, memory chips
- Massively parallel nature
- Well suited for VLSI
# Uniformity in Analysis
- Are domain agnostic in application
- Analysis methods are the same
- Can share theories, learning algorithms
# Neurobiological Analogy
- Design analogous to brain
- Already a demonstrable fault-tolerant, powerful, fast, parallel processor
- To slight changes
- Rotation of target in images
- Doppler shift in radar
- Network needs to be invariant to these transformations
# Invariance
1. Invariance by Structure
- Synaptic connections created so that transformed input produces same output
- Set same weight for neurons of some geometric relationship to image
- Same distance from centre e.g.
- Number of connections becomes prohibitively large
2. Invariance by Training
- Train on different views/transformations
- Take advantage of inherent pattern classification abilities
- Training for invariance for one object is not necessarily going to train other classes for invariance
- Extra load on network to do more training
- Exacerbated with high dimensionality
3. Invariant Feature Space
- Extract invariant features
- Use network as classifier
- Relieves burden on network to achieve invariance
- Complicated decision boundaries
- Number of features applied to network reduced
- Invariance ensured
- Required prior knowledge