April 2017 - Pharo-users - lists.pharo.org

Geocoder Services in Pharo?

by volkert

Dear all, do we have a package to access the a geocoder service? Something similar to "geopy". Volkert

April 29, 2017

Re: [Pharo-users] Neural Networks in Pharo

by Oleks

And I have added a PolyMath dependency to the configuration file, so now the Metacello script should update you to the latest stable versions of everything automatically. Thanks for pointing it out. On Fri, Apr 28, 2017 at 11:05 PM, francescoagati [via Smalltalk] < ml+s1294792n4944863h78(a)n4.nabble.com> wrote: > now work all good with last version of polymath > thanks :-) > > 2017-04-27 17:29 GMT+02:00 Oleks <[hidden email] > <http:///user/SendEmail.jtp?type=node&node=4944863&i=0>>: > >> Hello, >> >> I have finally added a configuration to the NeuralNetwork project. Now >> you can use this Metacello script to load it into your Pharo image: >> >> Metacello new >> repository: 'http://smalltalkhub.com/mc/Oleks/NeuralNetwork/main'; >> configuration: 'MLNeuralNetwork'; >> version: #development; >> load. >> >> Sorry for the delay >> >> Oleks >> >> On Tue, Apr 25, 2017 at 4:13 PM, francescoagati [via Smalltalk] <[hidden >> email] <http:///user/SendEmail.jtp?type=node&node=4944473&i=0>> wrote: >> >>> thanks ;-) >>> >>> 2017-04-25 15:09 GMT+02:00 Oleks <[hidden email] >>> <http:///user/SendEmail.jtp?type=node&node=4944028&i=0>>: >>> >>>> Hello, >>>> >>>> There isn't one yet. But I will try to create it today. I will let you >>>> know >>>> >>>> Cheers, >>>> Oleks >>>> >>>> On Apr 25, 2017 16:10, "francescoagati [via Smalltalk]" <[hidden email] >>>> <http:///user/SendEmail.jtp?type=node&node=4944027&i=0>> wrote: >>>> >>>>> Hi Oleks, >>>>> there is a mode for install neural network from metacello? >>>>> >>>>> 2017-04-25 13:00 GMT+02:00 Alexandre Bergel <[hidden email] >>>>> <http:///user/SendEmail.jtp?type=node&node=4944025&i=0>>: >>>>> >>>>>> Continue to push that topic Oleks. You are on the right track! >>>>>> >>>>>> Alexandre >>>>>> >>>>>> > On Apr 24, 2017, at 1:43 AM, Oleks <[hidden email] >>>>>> <http:///user/SendEmail.jtp?type=node&node=4944025&i=1>> wrote: >>>>>> > >>>>>> > Hello, >>>>>> > >>>>>> > Thanks a lot for your advice! It was very helpful and educating (for >>>>>> > example, I thought that we store biases in the weight matrix and >>>>>> prepend 1 >>>>>> > to input to make it faster, but now I see why it's actually slower >>>>>> that >>>>>> > way). >>>>>> > >>>>>> > I've implemented a multi-layer neural network as a linked list of >>>>>> layers >>>>>> > that propagate the input and error from one to another, similar to >>>>>> the Chain >>>>>> > of Responsibility pattern. Also, now I represent biases as separate >>>>>> vectors. >>>>>> > The LearningAlgorithm is a separate class with Backpropagation as >>>>>> its >>>>>> > subclass (though at this point the network can only learn through >>>>>> > backpropagation, but I'm planning to change that). I'm trying to >>>>>> figure out >>>>>> > how the activation and cost functions should be connected. For >>>>>> example, >>>>>> > cross-entropy works best with logistic sigmoid activation etc. I >>>>>> would like >>>>>> > to give the user a freedom to use whatever he wants (plug in >>>>>> whatever you >>>>>> > like and see what happens), but it can be very inefficient (because >>>>>> some >>>>>> > time-consuming parts of activation and cost derivatives cancel out >>>>>> each >>>>>> > other). >>>>>> > >>>>>> > Also, there is an interface for setting the learning rate for the >>>>>> whole >>>>>> > network, which can be used to choose the learning rate prior to >>>>>> learning, as >>>>>> > well as to change the learning rate after each iteration. I am >>>>>> planning to >>>>>> > implement some optimization algorithms that would automize the >>>>>> process of >>>>>> > choosing a learning rate (adagrad for example), but this would >>>>>> require a bit >>>>>> > different design (maybe I will implement the Optimizer, as you >>>>>> suggested). >>>>>> > >>>>>> > I'm attaching two images with UML diagrams, describing my current >>>>>> > implementation. Could you please tell me what you think about this >>>>>> design? >>>>>> > The first image is a class diagram that shows the whole >>>>>> architecture, and >>>>>> > the second one is a sequence diagram of backpropagation. >>>>>> > >>>>>> > mlnn.png <http://forum.world.st/file/n4943698/mlnn.png> >>>>>> > backprop.png <http://forum.world.st/file/n4943698/backprop.png> >>>>>> > >>>>>> > Sincerely yours, >>>>>> > Oleksandr >>>>>> > >>>>>> > >>>>>> > >>>>>> > -- >>>>>> > View this message in context: http://forum.world.st/Neural-N >>>>>> etworks-in-Pharo-tp4941271p4943698.html >>>>>> > Sent from the Pharo Smalltalk Users mailing list archive at >>>>>> Nabble.com. >>>>>> > >>>>>> >>>>>> -- >>>>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >>>>>> Alexandre Bergel http://www.bergel.eu >>>>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>>> >>>>> ------------------------------ >>>>> If you reply to this email, your message will be added to the >>>>> discussion below: >>>>> http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944025.html >>>>> To unsubscribe from Neural Networks in Pharo, click here. >>>>> NAML >>>>> <http://forum.world.st/template/NamlServlet.jtp?macro=macro_viewer&id=instan…> >>>>> >>>> >>>> ------------------------------ >>>> View this message in context: Re: Neural Networks in Pharo >>>> <http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944027.html> >>>> >>>> Sent from the Pharo Smalltalk Users mailing list archive >>>> <http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html> at >>>> Nabble.com. >>>> >>> >>> >>> >>> ------------------------------ >>> If you reply to this email, your message will be added to the discussion >>> below: >>> http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944028.html >>> To unsubscribe from Neural Networks in Pharo, click here. >>> NAML >>> <http://forum.world.st/template/NamlServlet.jtp?macro=macro_viewer&id=instan…> >>> >> >> >> ------------------------------ >> View this message in context: Re: Neural Networks in Pharo >> <http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944473.html> >> Sent from the Pharo Smalltalk Users mailing list archive >> <http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html> at >> Nabble.com. >> > > > > ------------------------------ > If you reply to this email, your message will be added to the discussion > below: > http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944863.html > To unsubscribe from Neural Networks in Pharo, click here > <http://forum.world.st/template/NamlServlet.jtp?macro=unsubscribe_by_code&no…> > . > NAML > <http://forum.world.st/template/NamlServlet.jtp?macro=macro_viewer&id=instan…> > -- View this message in context: http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944890.html Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com.

April 29, 2017

Re: [Pharo-users] Dark theme and white vs. green color

by john pfersich

Thanks, I was looking for "color" in an already open settings browser. That doesn't work. Sent from my iPhone > On Apr 28, 2017, at 18:15, Ben Coman <btc(a)openinworld.com> wrote: > > > >> On Sat, Apr 29, 2017 at 5:02 AM, john pfersich <jpfersich(a)gmail.com> wrote: >> I agree, that dark theme is harder to read than green on black. I just want the old light theme, how do I get it? >> >> Sent from my iPhone >> >> > On Apr 28, 2017, at 09:42, Juraj Kubelka <juraj.kubelka(a)icloud.com> wrote: >> > >> > it: > > Go to System > Settings and search for "theme". > > cheers -ben >

April 29, 2017

Re: [Pharo-users] Dark theme and white vs. green color

by Ben Coman

On Sat, Apr 29, 2017 at 5:02 AM, john pfersich <jpfersich(a)gmail.com> wrote: > I agree, that dark theme is harder to read than green on black. I just > want the old light theme, how do I get it? > > Sent from my iPhone > > > On Apr 28, 2017, at 09:42, Juraj Kubelka <juraj.kubelka(a)icloud.com> > wrote: > > > > it: > Go to System > Settings and search for "theme". cheers -ben

April 29, 2017

Re: [Pharo-users] Spec Adapter Advice

by Ben Coman

On Sat, Apr 29, 2017 at 4:20 AM, Rob Rothwell <r.j.rothwell(a)gmail.com> wrote: > Hi Peter, > > Thanks again for the advice. I've reduced my number of infinite loops and > crashes by creating copies of pertinent classes and using them for testing, > but I have quickly learned to save my image after inserting a halt anywhere! > Its a bit of a trick to learn to change the wheels on the car while its still driving, but also very useful. Try using #haltOnce instead. cheers -ben > > The parallel Spec implementation conversation is interesting too, but > unfortunately I don't understand enough to participate very much yet. > > However, I have discovered by creating new ComposableModel, WindowModel, > and MorphicWindowAdapter classes that even the SpecInterpreter itself is > hard-wired to use Morphic. > > If you take a look at > > WindowModel>>privateAdapterFromModel: aModel withSpec: aSpec > "apparently when looking at the implementation, it does not return a > widget but an adapter so it should be called adapter :)" > self halt. > ^ SpecInterpreter private_buildWidgetFor: self withSpec: aSpec. > > You see from the correct comment that an adapter is created during this > operation, which I think (I'm not that far yet) comes from > > SpecInterpreter class>>defaultBindings > ^ MorphicAdapterBindings new > > In other words, I think that regardless of WindowModel class>>adapterName > (and presumably all other "Model" classes, the actual adapter used is at > present determined by the SpecInterpreter via MorphicAdapterBindings>> > initializeBindings. > > You can set the bindings on the class side, but I currently don't see a > way to say "open this spec using these bindings." In other words, > something like: > > MyApplication new openWithSpecAndBindings: MorphicAdapterBindings. > > or simply > > MyApplication openAsMorphicSpec, MyApplication openAsCocoaSpec, etc... > > It's starting to make some sense at a high level, though. Eventually I'll > find a spot where I can create a CocoaWindow and start figuring out the > minimum interface requirements! > > Take care, > > Rob > > > > > > > On Fri, Apr 28, 2017 at 6:13 AM, Peter Uhnak <i.uhnak(a)gmail.com> wrote: > >> Hi Rob, >> >> > I need to find a good starting point because you can't poke at the >> existing >> > code too much since even the debugger is built with Spec! >> >> I don't know what Pharo you are on, but this is not the case. Debugger >> itself is based on Glamour (there used to be SpecDebugger, but it is no >> longer used). >> >> What is blowing up on you is the predebugger window, which is in Spec. >> You can disable this in your settings or by calling >> >> GTGenericStackDebugger alwaysOpenFullDebugger: true >> >> this will skip the debugger. >> >> But keep in mind that other tools might be using spec too (like the >> Browser window that opens when you are listing senders/implementors. >> >> overriding the #windowClass (or whatever the name was) is a good start. >> >> (also FML because apparently I've deleted my image with my spec changes >> by accident -_-) >> >> Peter >> >> > >> > It's pretty exciting, though, because it feels like all the right pieces >> > are there and a lot of this could mostly be "just work" once I get >> going. >> > >> > Thanks again, >> > >> > Rob >> > >> > On Thu, Apr 27, 2017 at 6:04 PM, Peter Uhnak <i.uhnak(a)gmail.com> wrote: >> > >> > > Hi Rob, >> > > >> > > I guess the best resource at the moment is to just look at existing >> > > adapters. >> > > >> > > But here are some thoughts to get you started: >> > > >> > > (Spec) ComposableModel has as its widget (Spec) Adapter >> > > Adapter has as its widget the actual visual component, which currently >> > > means Morphic. >> > > >> > > >> > > The simplest example: LabelModel >> > > >> > > 1. model to adapter >> > > >> > > LabelModel is adapted by MorphicLabelAdapter: >> > > >> > > you can see on the class side in method `defaultSpec`. >> > > There is only `LabelAdapter` there because there is some extra >> mapping for >> > > some classes, but you can specify directly the class of the Adapter, >> so the >> > > following would work just as well: >> > > >> > > LabelModel class>>defaultSpec >> > > <spec> >> > > >> > > ^ #(MorphicLabelAdapter >> > > adapt: #(model)) >> > > >> > > >> > > LabelModel -> LabelAdapter direct event propagation (pushing changes) >> > > >> > > in Model world you can call changed:with: to execute code on the >> Adapter, >> > > e.g. >> > > >> > > self changed: #emphasis: with: {emphasis} >> > > >> > > which in turn will call the #emphasis: method on the LabelAdapter. >> > > >> > > LabelModel -> LabelAdapter events (pulling changes) >> > > >> > > preferred alternative is to do the opposite; properties of models are >> > > often held in ValueHolders, which can be observed on ValueChange, so >> the >> > > Adapter can register to event change and react on itself; >> > > >> > > you can see this best in MorphicTabAdapter (but you will see that >> > > TabAdapter looks a bit different from the rest, because I was >> experimenting >> > > with cleaning up Adapters...) >> > > >> > > 2. adapter to morphic >> > > >> > > Every adapter implements #buildWidget method that returns a Morphic >> Object. >> > > Typically the adapter registers itself as the model of the adapter (so >> > > adapter's model is Spec, and Morphic's model is the adapter). This >> depends >> > > a lot on the API of morphic. (in TabAdapter I'm overriding adapt: so >> > > there's no buildWidget, and it is in #buildWidgetWith:). >> > > >> > > It is all kinds of messy, but if you have other widgets then it will >> be a >> > > good test how well Spec can handle it... or rather we'll see how it >> can be >> > > improved... >> > > >> > > Peter >> > > >> > > >> > > >> > > On Thu, Apr 27, 2017 at 05:10:39PM -0400, Rob Rothwell wrote: >> > > > Hello, >> > > > >> > > > I have recently been lucky enough to get the Cocoa portion of Mars >> > > working >> > > > on Pharo 5.0. While there are some issues, it has a wonderful set >> of >> > > > examples that allowed me create new Cocoa classes and delegate >> methods >> > > and >> > > > see a straightforward path to learning how to use Coco widgets from >> > > Pharo. >> > > > >> > > > I'd like to do that by creating appropriate Spec adapters, but >> > > > unfortunately the new Spec booklet wasn't designed for that! >> > > > >> > > > Can anyone give me any insight into creating (and using) a new Spec >> > > > adapter? Maybe just some high level steps to help me orient my >> > > > investigation into how Spec works? >> > > > >> > > > Thank you, >> > > > >> > > > Rob >> > > >> > > >> >> >

April 29, 2017

Re: [Pharo-users] Voyage - collecting data from Mongo

by Holger Freyther

> On 28. Apr 2017, at 14:27, Mark Rizun <mrizun(a)gmail.com> wrote: > > Hi, Hi! > Is it possible to retrieve data from Mongo collection if it was not created > via Voyage? > Meaning that I do not have a class in Pharo that would correspond to said > collection (should I implement one?). Yes. But you probably need to customize a bit. 1.) Make sure you refer to the right collection... descriptionContainer <mongoContainer> ^VOMongoContainer new collectionName: 'yourExistingCollectionName' yourself 2.) For materialization.. you should at least have _id (okay every entry does that) VOMongoSerializer fieldVersion (#version on my old Voyage) VOMongoSerializer fieldType (#instanceOf) and then things should work out. You can probably add #version and #instanceOf to your existing objects? holger

April 28, 2017

Re: [Pharo-users] Dark theme and white vs. green color

by john pfersich

I agree, that dark theme is harder to read than green on black. I just want the old light theme, how do I get it? Sent from my iPhone > On Apr 28, 2017, at 09:42, Juraj Kubelka <juraj.kubelka(a)icloud.com> wrote: > > it:

April 28, 2017

Re: [Pharo-users] Spec Adapter Advice

by Rob Rothwell

Hi Peter, Thanks again for the advice. I've reduced my number of infinite loops and crashes by creating copies of pertinent classes and using them for testing, but I have quickly learned to save my image after inserting a halt anywhere! The parallel Spec implementation conversation is interesting too, but unfortunately I don't understand enough to participate very much yet. However, I have discovered by creating new ComposableModel, WindowModel, and MorphicWindowAdapter classes that even the SpecInterpreter itself is hard-wired to use Morphic. If you take a look at WindowModel>>privateAdapterFromModel: aModel withSpec: aSpec "apparently when looking at the implementation, it does not return a widget but an adapter so it should be called adapter :)" self halt. ^ SpecInterpreter private_buildWidgetFor: self withSpec: aSpec. You see from the correct comment that an adapter is created during this operation, which I think (I'm not that far yet) comes from SpecInterpreter class>>defaultBindings ^ MorphicAdapterBindings new In other words, I think that regardless of WindowModel class>>adapterName (and presumably all other "Model" classes, the actual adapter used is at present determined by the SpecInterpreter via MorphicAdapterBindings>>initializeBindings. You can set the bindings on the class side, but I currently don't see a way to say "open this spec using these bindings." In other words, something like: MyApplication new openWithSpecAndBindings: MorphicAdapterBindings. or simply MyApplication openAsMorphicSpec, MyApplication openAsCocoaSpec, etc... It's starting to make some sense at a high level, though. Eventually I'll find a spot where I can create a CocoaWindow and start figuring out the minimum interface requirements! Take care, Rob On Fri, Apr 28, 2017 at 6:13 AM, Peter Uhnak <i.uhnak(a)gmail.com> wrote: > Hi Rob, > > > I need to find a good starting point because you can't poke at the > existing > > code too much since even the debugger is built with Spec! > > I don't know what Pharo you are on, but this is not the case. Debugger > itself is based on Glamour (there used to be SpecDebugger, but it is no > longer used). > > What is blowing up on you is the predebugger window, which is in Spec. You > can disable this in your settings or by calling > > GTGenericStackDebugger alwaysOpenFullDebugger: true > > this will skip the debugger. > > But keep in mind that other tools might be using spec too (like the > Browser window that opens when you are listing senders/implementors. > > overriding the #windowClass (or whatever the name was) is a good start. > > (also FML because apparently I've deleted my image with my spec changes by > accident -_-) > > Peter > > > > > It's pretty exciting, though, because it feels like all the right pieces > > are there and a lot of this could mostly be "just work" once I get going. > > > > Thanks again, > > > > Rob > > > > On Thu, Apr 27, 2017 at 6:04 PM, Peter Uhnak <i.uhnak(a)gmail.com> wrote: > > > > > Hi Rob, > > > > > > I guess the best resource at the moment is to just look at existing > > > adapters. > > > > > > But here are some thoughts to get you started: > > > > > > (Spec) ComposableModel has as its widget (Spec) Adapter > > > Adapter has as its widget the actual visual component, which currently > > > means Morphic. > > > > > > > > > The simplest example: LabelModel > > > > > > 1. model to adapter > > > > > > LabelModel is adapted by MorphicLabelAdapter: > > > > > > you can see on the class side in method `defaultSpec`. > > > There is only `LabelAdapter` there because there is some extra mapping > for > > > some classes, but you can specify directly the class of the Adapter, > so the > > > following would work just as well: > > > > > > LabelModel class>>defaultSpec > > > <spec> > > > > > > ^ #(MorphicLabelAdapter > > > adapt: #(model)) > > > > > > > > > LabelModel -> LabelAdapter direct event propagation (pushing changes) > > > > > > in Model world you can call changed:with: to execute code on the > Adapter, > > > e.g. > > > > > > self changed: #emphasis: with: {emphasis} > > > > > > which in turn will call the #emphasis: method on the LabelAdapter. > > > > > > LabelModel -> LabelAdapter events (pulling changes) > > > > > > preferred alternative is to do the opposite; properties of models are > > > often held in ValueHolders, which can be observed on ValueChange, so > the > > > Adapter can register to event change and react on itself; > > > > > > you can see this best in MorphicTabAdapter (but you will see that > > > TabAdapter looks a bit different from the rest, because I was > experimenting > > > with cleaning up Adapters...) > > > > > > 2. adapter to morphic > > > > > > Every adapter implements #buildWidget method that returns a Morphic > Object. > > > Typically the adapter registers itself as the model of the adapter (so > > > adapter's model is Spec, and Morphic's model is the adapter). This > depends > > > a lot on the API of morphic. (in TabAdapter I'm overriding adapt: so > > > there's no buildWidget, and it is in #buildWidgetWith:). > > > > > > It is all kinds of messy, but if you have other widgets then it will > be a > > > good test how well Spec can handle it... or rather we'll see how it > can be > > > improved... > > > > > > Peter > > > > > > > > > > > > On Thu, Apr 27, 2017 at 05:10:39PM -0400, Rob Rothwell wrote: > > > > Hello, > > > > > > > > I have recently been lucky enough to get the Cocoa portion of Mars > > > working > > > > on Pharo 5.0. While there are some issues, it has a wonderful set of > > > > examples that allowed me create new Cocoa classes and delegate > methods > > > and > > > > see a straightforward path to learning how to use Coco widgets from > > > Pharo. > > > > > > > > I'd like to do that by creating appropriate Spec adapters, but > > > > unfortunately the new Spec booklet wasn't designed for that! > > > > > > > > Can anyone give me any insight into creating (and using) a new Spec > > > > adapter? Maybe just some high level steps to help me orient my > > > > investigation into how Spec works? > > > > > > > > Thank you, > > > > > > > > Rob > > > > > > > >

April 28, 2017

Re: [Pharo-users] Neural Networks in Pharo

by francesco agati

now work all good with last version of polymath thanks :-) 2017-04-27 17:29 GMT+02:00 Oleks <olk.zaytsev(a)gmail.com>: > Hello, > > I have finally added a configuration to the NeuralNetwork project. Now you > can use this Metacello script to load it into your Pharo image: > > Metacello new > repository: 'http://smalltalkhub.com/mc/Oleks/NeuralNetwork/main'; > configuration: 'MLNeuralNetwork'; > version: #development; > load. > > Sorry for the delay > > Oleks > > On Tue, Apr 25, 2017 at 4:13 PM, francescoagati [via Smalltalk] <[hidden > email] <http:///user/SendEmail.jtp?type=node&node=4944473&i=0>> wrote: > >> thanks ;-) >> >> 2017-04-25 15:09 GMT+02:00 Oleks <[hidden email] >> <http:///user/SendEmail.jtp?type=node&node=4944028&i=0>>: >> >>> Hello, >>> >>> There isn't one yet. But I will try to create it today. I will let you >>> know >>> >>> Cheers, >>> Oleks >>> >>> On Apr 25, 2017 16:10, "francescoagati [via Smalltalk]" <[hidden email] >>> <http:///user/SendEmail.jtp?type=node&node=4944027&i=0>> wrote: >>> >>>> Hi Oleks, >>>> there is a mode for install neural network from metacello? >>>> >>>> 2017-04-25 13:00 GMT+02:00 Alexandre Bergel <[hidden email] >>>> <http:///user/SendEmail.jtp?type=node&node=4944025&i=0>>: >>>> >>>>> Continue to push that topic Oleks. You are on the right track! >>>>> >>>>> Alexandre >>>>> >>>>> > On Apr 24, 2017, at 1:43 AM, Oleks <[hidden email] >>>>> <http:///user/SendEmail.jtp?type=node&node=4944025&i=1>> wrote: >>>>> > >>>>> > Hello, >>>>> > >>>>> > Thanks a lot for your advice! It was very helpful and educating (for >>>>> > example, I thought that we store biases in the weight matrix and >>>>> prepend 1 >>>>> > to input to make it faster, but now I see why it's actually slower >>>>> that >>>>> > way). >>>>> > >>>>> > I've implemented a multi-layer neural network as a linked list of >>>>> layers >>>>> > that propagate the input and error from one to another, similar to >>>>> the Chain >>>>> > of Responsibility pattern. Also, now I represent biases as separate >>>>> vectors. >>>>> > The LearningAlgorithm is a separate class with Backpropagation as its >>>>> > subclass (though at this point the network can only learn through >>>>> > backpropagation, but I'm planning to change that). I'm trying to >>>>> figure out >>>>> > how the activation and cost functions should be connected. For >>>>> example, >>>>> > cross-entropy works best with logistic sigmoid activation etc. I >>>>> would like >>>>> > to give the user a freedom to use whatever he wants (plug in >>>>> whatever you >>>>> > like and see what happens), but it can be very inefficient (because >>>>> some >>>>> > time-consuming parts of activation and cost derivatives cancel out >>>>> each >>>>> > other). >>>>> > >>>>> > Also, there is an interface for setting the learning rate for the >>>>> whole >>>>> > network, which can be used to choose the learning rate prior to >>>>> learning, as >>>>> > well as to change the learning rate after each iteration. I am >>>>> planning to >>>>> > implement some optimization algorithms that would automize the >>>>> process of >>>>> > choosing a learning rate (adagrad for example), but this would >>>>> require a bit >>>>> > different design (maybe I will implement the Optimizer, as you >>>>> suggested). >>>>> > >>>>> > I'm attaching two images with UML diagrams, describing my current >>>>> > implementation. Could you please tell me what you think about this >>>>> design? >>>>> > The first image is a class diagram that shows the whole >>>>> architecture, and >>>>> > the second one is a sequence diagram of backpropagation. >>>>> > >>>>> > mlnn.png <http://forum.world.st/file/n4943698/mlnn.png> >>>>> > backprop.png <http://forum.world.st/file/n4943698/backprop.png> >>>>> > >>>>> > Sincerely yours, >>>>> > Oleksandr >>>>> > >>>>> > >>>>> > >>>>> > -- >>>>> > View this message in context: http://forum.world.st/Neural-N >>>>> etworks-in-Pharo-tp4941271p4943698.html >>>>> > Sent from the Pharo Smalltalk Users mailing list archive at >>>>> Nabble.com. >>>>> > >>>>> >>>>> -- >>>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >>>>> Alexandre Bergel http://www.bergel.eu >>>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >>>>> >>>>> >>>>> >>>>> >>>>> >>>> >>>> >>>> ------------------------------ >>>> If you reply to this email, your message will be added to the >>>> discussion below: >>>> http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944025.html >>>> To unsubscribe from Neural Networks in Pharo, click here. >>>> NAML >>>> <http://forum.world.st/template/NamlServlet.jtp?macro=macro_viewer&id=instan…> >>>> >>> >>> ------------------------------ >>> View this message in context: Re: Neural Networks in Pharo >>> <http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944027.html> >>> >>> Sent from the Pharo Smalltalk Users mailing list archive >>> <http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html> at >>> Nabble.com. >>> >> >> >> >> ------------------------------ >> If you reply to this email, your message will be added to the discussion >> below: >> http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944028.html >> To unsubscribe from Neural Networks in Pharo, click here. >> NAML >> <http://forum.world.st/template/NamlServlet.jtp?macro=macro_viewer&id=instan…> >> > > > ------------------------------ > View this message in context: Re: Neural Networks in Pharo > <http://forum.world.st/Neural-Networks-in-Pharo-tp4941271p4944473.html> > Sent from the Pharo Smalltalk Users mailing list archive > <http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html> at Nabble.com. >

April 28, 2017

Re: [Pharo-users] Dark theme and white vs. green color

by Offray Vladimir Luna CÃ¡rdenas

I think that green reads better. Something like solarized will be really nice [1]. But maybe that is a lot of work now. [1] http://ethanschoonover.com/solarized Cheers, Offray On 28/04/17 11:42, Juraj Kubelka wrote: > Hi, > > While we were improving the DiffMorph for the dark theme, we conclude > that it is much harder to read the text. > > I personally also often miss a new debugger window on a big screen. > Something I have not experienced before. > > And this discussion reminds me the old computers where the background > was black and text green. There are two screenshot to compare it: > > > A personally feel that it is much easier to read the green text. > > What do you think? > Juraj >

April 28, 2017