***
But we need "fast and easy to create, destroy, change, highly mutable,
but searchable" entities. So, this is not only technical, but also
conceptual problem...
I would really like to hear your opinion on this. What should we do?
***
Hmm, well when I think about the algorithms involved, I do n
> One question is: Is probabilistic logic an appropriate method for the
> core of an AGI system, given that this AGI system must proceed largely
> on observation-based semantics ...
>
> I think the answer is YES
>
> Another question is: Is the current OpenCog infrastructure fully ready
> to support
Hi Alexey,
these are valid points. Currently, as you probably already understand,
the (only?) way to match values is to resort to grounded schemata, see
for instance
https://github.com/opencog/opencog/blob/ea987668ed713c55c2df087b81f55736d7469772/opencog/learning/miner/rules/shallow-abstracti
>
> Hmm, well when I think about the algorithms involved, I do not see why
> the Pattern Miner and Pattern Matcher would be unable to search for
> patterns involving Values... I think they could It's true the
> code doesn't do this now though...
>
Yes, it should be quite possible algorithmica
Hi Nil,
Currently, as you probably already understand, the (only?) way to match
> values is to resort to grounded schemata.
>
Conceptually, this might be ok. We have conscious processes implemented in
a self-reflective Atomese and operating over atoms, and subconscious
processes operating over va
Hi Alexey,
I will answer this email in several parts. Re: atoms vs values, my
thinking is this:
-- Use Atoms to represent the "topology" of a network: what is connected to
what. Atoms express (long-term, slowly-varying) relationships between
things.
-- Use Values to hold fast-changing data. F
Linas,
>
> I will answer this email in several parts. Re: atoms vs values, my
> thinking is this:
>
> -- Use Atoms to represent the "topology" of a network: what is connected
> to what. Atoms express (long-term, slowly-varying) relationships between
> things.
>
> -- Use Values to hold fast-chan
Ongoing breakup of TL;DR into small pieces.
On Sun, May 20, 2018 at 10:53 AM, Alexey Potapov wrote:
>
> Both tasks can be considered as a part of the Semantic Vision problem, but
> their solution can be useful in a more general context.
>
> *OpenCog + Tensorflow*
> Depth of OpenCog+Tensoflow int
On Sun, May 20, 2018 at 4:45 PM, Ben Goertzel wrote:
>
>
> Hmm, well when I think about the algorithms involved, I do not see why
> the Pattern Miner and Pattern Matcher would be unable to search for
> patterns involving Values...
>
Here is the key difference between the two: Atoms know what the
On Mon, May 21, 2018 at 3:40 PM, Alexey Potapov wrote:
>
> Yes, this is clear (although I'd like to know more about your ideas
> regarding connectors),
>
Sure; say when. I can talk about them for days. Unfortunately, the core
idea is so simple, so obvious, that it becomes very difficult to talk
On 05/21/2018 10:51 PM, Alexey Potapov wrote:
processes operating over values in a native code. However, what we lack
in this case is meta-computations: specialization of a conscious
decision-making w.r.t. some specific task should yield its efficient
implementation in a native code (or trained
On 05/21/2018 11:11 PM, Linas Vepstas wrote:
actually make the atom. Atoms are also held in an index, so that they
can be searched by name, by type. Insertion into an index is expensive
-- and stupid, if you never use the index. Values avoid this overhead.
Could this indexing be made lazy?
On 05/22/2018 08:34 AM, Nil Geisweiller wrote:
Do you mean that this link is needed for Pattern Matcher to explicitly
know that we want to apply TimesLink, etc. to Values? I guess, it
makes sense, although this might be not too efficient, indeed.
The atomization might be the problem, but if it
Linas,
>
>> Yes, this is clear (although I'd like to know more about your ideas
>> regarding connectors),
>>
>
> Sure; say when. I can talk about them for days. Unfortunately, the core
> idea is so simple, so obvious, that it becomes very difficult to talk about
> the advanced concepts, so this
Nil,
>
> What seems to happen in human is that high confidence knowledge tend
> to go subconscious, while less confidence knowledge tend to be the
> subject of the attention. So for instance a at first you may focus on
> edge detection, etc, once you've built some high confidence model
> about th
Alexey,
On 05/23/2018 03:44 PM, Alexey Potapov wrote:
Nil,
What seems to happen in human is that high confidence knowledge tend
to go subconscious, while less confidence knowledge tend to be the
subject of the attention. So for instance a at first you may focus on
edge detectio
On Tue, May 22, 2018 at 12:34 AM, Nil Geisweiller
wrote:
>
>> What do you suppose GetValueLink to do?
>>
>
> For instance
>
> GetValueLink
>
>
>
> would return the value corresponding to `key` in `atom`. Note that
> `key` is itself an atom. However the returned value may not
> necessarily
On Tue, May 22, 2018 at 12:48 AM, Nil Geisweiller
wrote:
> On 05/21/2018 11:11 PM, Linas Vepstas wrote:
>
>> actually make the atom. Atoms are also held in an index, so that they
>> can be searched by name, by type. Insertion into an index is expensive --
>> and stupid, if you never use the inde
On Tue, May 22, 2018 at 1:05 AM, Nil Geisweiller
wrote:
> On 05/22/2018 08:34 AM, Nil Geisweiller wrote:
>
>> Do you mean that this link is needed for Pattern Matcher to explicitly
>>> know that we want to apply TimesLink, etc. to Values? I guess, it makes
>>> sense, although this might be not to
On Wed, May 23, 2018 at 7:19 AM, Alexey Potapov wrote:
> Linas,
>
>
>
>>
>>> Yes, this is clear (although I'd like to know more about your ideas
>>> regarding connectors),
>>>
>>
>> Sure; say when. I can talk about them for days. Unfortunately, the core
>> idea is so simple, so obvious, that it
On Wed, May 23, 2018 at 9:00 AM, Nil Geisweiller
wrote:
> I mean when run by cog-execute!, which also happens to be the standard
way of invoking the pattern matcher.
Any atom type that derives from FunctionLink has the C++ method
virtual Handle execute() const;
method on it ... which cog
On 05/24/2018 07:24 AM, Linas Vepstas wrote:
Could this indexing be made lazy?
No, its fundamental to what the definition of the atomspace is. Its the
only way that you can have a single, unique (Concept "cat") in the
system.
But the notion of uniqueness is only relevant when you query so
On Thu, May 24, 2018 at 1:18 AM, Nil Geisweiller
wrote:
> On 05/24/2018 07:24 AM, Linas Vepstas wrote:
>
>> Could this indexing be made lazy?
>>
>>
>> No, its fundamental to what the definition of the atomspace is. Its the
>> only way that you can have a single, unique (Concept "cat") in the
2018-05-22 0:11 GMT+03:00 Linas Vepstas :
>
> How many are we talking about, here? dozens, hundreds of objects? hundreds
> of predicates per object? That is 100x100 = 10K and, currently, you can
> create and add maybe 100K atoms/sec to the atomspace (via C++, less by
> scheme, python, due to wrapp
Hi Alexey,
> Yes, it should be quite possible algorithmically. And that's exactly why we
> discuss this - because we want to use PM algorithms on Values. However, to
> implement this, some architectural and organizational decisions should be
> made (should we generalize existing values to tensors
On 05/22/2018 02:11 AM, Alexey Potapov wrote:
Perhaps we need an IsLeftOfLink that knows automatically to obtain
the "*-centroid-*" value on two atoms, and then return true/false
depending on the result (or throw exception if there is no
*-centroid-* value.)
Sorry, I did't not p
Hi Ben,
> I think it's better, if possible, to figure out a way to suitably
> modify the core PM rather than using a separate repo ...
>
> However, I guess the PM tweaks would need to be done someone on your team,
> as
> Linas and Nil probably are too busy and we don't have a lot of others
> who
> I have also an idea about using evolutionary programming for training
> combined Atomese/DNN models (e.g. for SynerGAN-ish models or VQA)...
Ah, interesting.
Yes, an advantage of evolutionary algorithms is that they apply to
basically any data type, or any combination thereof, etc. ..
They ar
On Mon, May 21, 2018 at 6:11 PM, Alexey Potapov wrote:
>
>
> 2018-05-22 0:11 GMT+03:00 Linas Vepstas :
>
>>
>> How many are we talking about, here? dozens, hundreds of objects?
>> hundreds of predicates per object? That is 100x100 = 10K and, currently,
>> you can create and add maybe 100K atoms/s
On 05/24/2018 06:33 AM, Linas Vepstas wrote:
> Nil also proposed to use GetValueLink...
I didn't really understand that proposal. He seemed to be talking about
truth values, not values in general.
No, I meant values in general. So for instance if atom (Node "A") holds
a FloatValue (5.2, 0.1
Hi Ben,
I'm trying to spend just enough time to transmit all of the key design
principles and constraints to Alexey and his team - (and Anton and his
team) - as they are first-in-line to then diseminate this knowledge to a
wider group. Transmitting this information accurately is very key, very
im
On Tue, May 22, 2018 at 9:26 PM, Alexey Potapov wrote:
>
>
>
>>
>> If needed we could also introduce some sort of entity that is between
>> a Value and an Atom in some sense -- i.e. we could introduce some sort of
>> TensorValue entity that
>>
>> 1) Perhaps, knows what links to it (like an Atom b
***
The existing architecture has room for a lot of things, a lot of
freedom for designing things. I'd like to stick to it as much as
possible.
***
It's very understandable, however "... as possible" is key here, and
it's hard to see how the current system can scalably deal with tensors
from senso
On Thu, May 24, 2018 at 1:08 AM, Ben Goertzel wrote:
> ***
> The existing architecture has room for a lot of things, a lot of
> freedom for designing things. I'd like to stick to it as much as
> possible.
> ***
>
> It's very understandable, however "... as possible" is key here, and
> it's hard t
Linas,
> I want to keep this conversation realistic. Sophia, today, struggles to
> see human faces.
>
We don't talk about applying existing narrow method. These method may be
needed/realistic/practical now, but they don't bring us much closer to AGI.
If we were talking about 'realistic' things
>
>
>>
> No, I meant values in general. So for instance if atom (Node "A") holds a
> FloatValue (5.2, 0.1, 4.5) at key (Schema "*-my-key*-"), the following
> would return something like
>
> (cog-execute!
> (GetValueLink
> (Node "A")
> (Schema "*-my-key-*")))
>
> (List (Number 5.2) (Number
On Thu, May 24, 2018 at 6:03 AM, Alexey Potapov wrote:
> Linas,
>
>
>> I want to keep this conversation realistic. Sophia, today, struggles to
>> see human faces.
>>
>
> We don't talk about applying existing narrow method. These method may be
> needed/realistic/practical now, but they don't brin
On Thu, May 24, 2018 at 6:13 AM, Alexey Potapov wrote:
>
>>>
>> No, I meant values in general. So for instance if atom (Node "A") holds a
>> FloatValue (5.2, 0.1, 4.5) at key (Schema "*-my-key*-"), the following
>> would return something like
>>
>> (cog-execute!
>> (GetValueLink
>> (Node "A
2018-05-24 8:42 GMT+03:00 Linas Vepstas :
>
> Actually, I want you to not think about this. I strongly believe that
> pretty much anything you can think of will fit nicely into an Atom, or into
> a Value. I do not want to see a third kind of "generic object system"
> being created, that would be
Hi.
Here are some additional thoughts on OpenCog+PPL, which I didn't include in
the first message
*OpenCog + PPL*
One of the ideas/tasks was to make an OpenCog a “better Church”.
*1. **Why is it possible*
Universal probabilistic programming languages (PPLs) utilize sampling-based
approa
On Mon, May 28, 2018 at 8:27 AM, Alexey Potapov wrote:
> Hi.
> Here are some additional thoughts on OpenCog+PPL, which I didn't include
> in the first message
>
> *OpenCog + PPL*
>
>
>
> One of the ideas/tasks was to make an OpenCog a “better Church”.
>
Oooof. I wrote a long diatribe about relig
>
>
> Oooof. I wrote a long diatribe about religion, until I realized you must
> be referring to something that Alonzo Church must have written or invented.
> I am not that familiar with his work, so I don't know what a "better
> Church" would be. Is there some specific paper or book?
>
https://
42 matches
Mail list logo
