Golang rlglue.Observation类代码示例

原作者: [db:作者] 来自: [db:来源] 收藏邀请

本文整理汇总了Golang中go-glue/googlecode/com/hg/rlglue.Observation类的典型用法代码示例。如果您正苦于以下问题：Golang Observation类的具体用法？Golang Observation怎么用？Golang Observation使用的例子？那么恭喜您, 这里精选的类代码示例或许可以为您提供帮助。

在下文中一共展示了Observation类的8个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于我们的系统推荐出更棒的Golang代码示例。

示例1: AgentStart

func (ra *RmaxFSSSAgent) AgentStart(obs rlglue.Observation) (act rlglue.Action) {
	ra.stepsWithPlanner = 0
	ra.lastState = discrete.State(ra.task.Obs.Ints.Index(obs.Ints()))
	ra.Plan()
	act = rlglue.NewAction(ra.task.Act.Ints.Values(ra.GetAction().Hashcode()), []float64{}, []byte{})
	ra.lastAction = discrete.Action(ra.task.Act.Ints.Index(act.Ints()))
	return
}

开发者ID:skelterjohn，项目名称:rlalg，代码行数:8，代码来源:rmaxfs3.go

示例2: AgentStart

func (this *Agent) AgentStart(obs rlglue.Observation) (act rlglue.Action) {
	this.stepsWithPlanner = 0
	this.lastState = discrete.State(this.mdp.GetTask().Obs.Ints.Index(obs.Ints()))
	this.Plan()
	act = rlglue.NewAction(this.mdp.GetTask().Act.Ints.Values(this.GetAction()), []float64{}, []byte{})
	this.lastAction = discrete.Action(this.mdp.GetTask().Act.Ints.Index(act.Ints()))
	return
}

开发者ID:skelterjohn，项目名称:rlalg，代码行数:8，代码来源:fsssmdp.go

示例3: AgentStep

func (ra *BebAgent) AgentStep(reward float64, obs rlglue.Observation) (act rlglue.Action) {
	nextState := discrete.State(ra.task.Obs.Ints.Index(obs.Ints()))
	learned := ra.rmdp.Observe(ra.lastState, ra.lastAction, nextState, reward)
	if learned {
		vi.ValueIteration(ra.qt, ra.rmdp, ra.Cfg.Epsilon)
	}
	ra.lastState = nextState
	act = rlglue.NewAction(ra.task.Act.Ints.Values(ra.qt.Pi(ra.lastState).Hashcode()), []float64{}, []byte{})
	ra.lastAction = discrete.Action(ra.task.Act.Ints.Index(act.Ints()))
	return
}

开发者ID:skelterjohn，项目名称:rlalg，代码行数:11，代码来源:beb.go

示例4: AgentStep

func (this *ROARAgent) AgentStep(reward float64, obs rlglue.Observation) rlglue.Action {
	last := matrix.MakeDenseMatrix(this.LastObs.Doubles(), this.numFeatures, 1)
	current := matrix.MakeDenseMatrix(obs.Doubles(), this.numFeatures, 1)
	rm := matrix.MakeDenseMatrix([]float64{reward}, 1, 1)
	outcome, _ := current.MinusDense(last)
	sor, _ := last.Augment(outcome)
	sor, _ = sor.Augment(rm)
	actionIndex := this.task.Act.Ints.Index(this.LastAct.Ints())
	this.rpost[actionIndex].Insert(sor)
	this.LastObs = obs
	return this.GetAction()
}

开发者ID:postfix，项目名称:rlbayes，代码行数:12，代码来源:agent.go

示例5: AgentStep

func (ra *RmaxFSSSAgent) AgentStep(reward float64, obs rlglue.Observation) (act rlglue.Action) {
	ra.stepsWithPlanner++
	nextState := discrete.State(ra.task.Obs.Ints.Index(obs.Ints()))
	learned := ra.rmdp.Observe(ra.lastState, ra.lastAction, nextState, reward)
	if learned {
		ra.Forget()
	}
	ra.lastState = nextState
	ra.Plan()
	act = rlglue.NewAction(ra.task.Act.Ints.Values(ra.GetAction().Hashcode()), []float64{}, []byte{})
	ra.lastAction = discrete.Action(ra.task.Act.Ints.Index(act.Ints()))
	return
}

开发者ID:skelterjohn，项目名称:rlalg，代码行数:13，代码来源:rmaxfs3.go

示例6: AgentStart

func (ra *BebAgent) AgentStart(obs rlglue.Observation) (act rlglue.Action) {
	ra.lastState = discrete.State(ra.task.Obs.Ints.Index(obs.Ints()))
	act = rlglue.NewAction(ra.task.Act.Ints.Values(ra.qt.Pi(ra.lastState).Hashcode()), []float64{}, []byte{})
	ra.lastAction = discrete.Action(ra.task.Act.Ints.Index(act.Ints()))
	return
}

开发者ID:skelterjohn，项目名称:rlalg，代码行数:6，代码来源:beb.go

示例7: getStateIndex

func (this *BFS3Agent) getStateIndex(state rlglue.Observation) (index uint64) {
	return this.task.Obs.Ints.Index(state.Ints())
}

开发者ID:skelterjohn，项目名称:rlalg，代码行数:3，代码来源:bfs3.go

示例8: AgentStep

func (this *OptAgent) AgentStep(reward float64, obs rlglue.Observation) (act rlglue.Action) {
	s := discrete.State(this.task.Obs.Ints.Index(obs.Ints()))
	a := this.qt.Pi(s)
	act = rlglue.NewAction([]int32{int32(a)}, []float64{}, []byte{})
	return
}

开发者ID:skelterjohn，项目名称:rlenv，代码行数:6，代码来源:opt.go

注：本文中的go-glue/googlecode/com/hg/rlglue.Observation类示例由纯净天空整理自Github/MSDocs等源码及文档管理平台，相关代码片段筛选自各路编程大神贡献的开源项目，源码版权归原作者所有，传播和使用请参考对应项目的License；未经允许，请勿转载。