agent.md

ReImproveJS > Agent

Class: Agent

Hierarchy

Agent

Index

Constructors

• constructor

Properties

• actionsBuffer
• agentConfig
• currentReward
• done
• forwardPasses
• inputsBuffer
• lossesHistory
• memory
• model
• name
• netInputWindowSize
• rewardsHistory
• statesBuffer

Accessors

• Config
• Name

Methods

• addReward
• createNeuralNetInput
• createTrainingDataFromMemento
• forward
• getTrackingInformation
• learn
• listen
• memorize
• policy
• reset
• setReward

---

Constructors

constructor

⊕ new Agent(model: Model, agentConfig?: AgentConfig, name?: string): Agent

Defined in reimprove/agent.ts:44

Parameters:

Param	Type
model	Model
Optional agentConfig	AgentConfig
Optional name	string

Returns: Agent

---

Properties

<Private> actionsBuffer

● actionsBuffer: Array<number>

Defined in reimprove/agent.ts:33

---

<Private> agentConfig

● agentConfig: AgentConfig

Defined in reimprove/agent.ts:42

---

<Private> currentReward

● currentReward: number

Defined in reimprove/agent.ts:31

---

<Private> done

● done: boolean

Defined in reimprove/agent.ts:30

---

<Private> forwardPasses

● forwardPasses: number

Defined in reimprove/agent.ts:44

---

<Private> inputsBuffer

● inputsBuffer: Array<MementoTensor>

Defined in reimprove/agent.ts:35

---

<Private> lossesHistory

● lossesHistory: TypedWindow<number>

Defined in reimprove/agent.ts:37

---

<Private> memory

● memory: Memory

Defined in reimprove/agent.ts:41

---

<Private> model

● model: Model

Defined in reimprove/agent.ts:47

---

<Private>``<Optional> name

● name: string

Defined in reimprove/agent.ts:47

---

<Private> netInputWindowSize

● netInputWindowSize: number

Defined in reimprove/agent.ts:39

---

<Private> rewardsHistory

● rewardsHistory: TypedWindow<number>

Defined in reimprove/agent.ts:38

---

<Private> statesBuffer

● statesBuffer: Array<Tensor>

Defined in reimprove/agent.ts:34

---

Accessors

Config

getConfig(): AgentConfig

Defined in reimprove/agent.ts:209

Returns: AgentConfig

---

Name

getName(): stringsetName(name: string): void

Defined in reimprove/agent.ts:217

Returns: string

Defined in reimprove/agent.ts:213

Parameters:

Param	Type
name	string

Returns: void

---

Methods

addReward

▸ addReward(value: number): void

Defined in reimprove/agent.ts:194

Parameters:

Param	Type
value	number

Returns: void

---

<Private> createNeuralNetInput

▸ createNeuralNetInput(input: Tensor): Tensor

Defined in reimprove/agent.ts:65

Parameters:

Param	Type
input	Tensor

Returns: Tensor

---

createTrainingDataFromMemento

▸ createTrainingDataFromMemento(memento: Memento, gamma: number): object

Defined in reimprove/agent.ts:145

Parameters:

Param	Type
memento	Memento
gamma	number

Returns: object

---

forward

▸ forward(input: number[], epsilon: number, keepTensors?: boolean): number

Defined in reimprove/agent.ts:90

Parameters:

Param	Type	Default value
input	number[]	-
epsilon	number	-
Default value keepTensors	boolean	true

Returns: number

---

getTrackingInformation

▸ getTrackingInformation(): AgentTrackingInformation

Defined in reimprove/agent.ts:221

Returns: AgentTrackingInformation

---

learn

▸ learn(gamma: number): Promise<void>

Defined in reimprove/agent.ts:167

Parameters:

Param	Type
gamma	number

Returns: Promise<void>

---

listen

▸ listen(input: number[], epsilon: number): number

Defined in reimprove/agent.ts:158

Parameters:

Param	Type
input	number[]
epsilon	number

Returns: number

---

memorize

▸ memorize(): void

Defined in reimprove/agent.ts:130

Returns: void

---

<Private> policy

▸ policy(input: Tensor): number

Defined in reimprove/agent.ts:86

Parameters:

Param	Type
input	Tensor

Returns: number

---

reset

▸ reset(): void

Defined in reimprove/agent.ts:202

Returns: void

---

setReward

▸ setReward(value: number): void

Defined in reimprove/agent.ts:198

Parameters:

Param	Type
value	number

Returns: void

---