Thank you Matt for your time and consideration. I always enjoy your comments and contributions to the group.
AGI Brain is still a newborn project and it is undergoing tests and gradual development in computational environments yet. The problems and environments that you stated (as well as other aspects of intelligence) seem to be complex problems to the model at this stage, but are planned to be the next developmental goals of the project, once the project is fully operational in a reasonable number of computational environments. I believe in gradual development of the model and for the near future, its developmental goals are delayed reward environments and the environments with intelligent opponents like games. Once these goals are achieved, I will work on perception (hearing, and then vision) first and then I want to deeply learn current powerful and general purpose models like NARS and OpenCog in order to include them in my model or vice versa. I am learning and studying proposed general purpose models at the for example AGI conferences to find the best propsed models that would help in developing AGI Brain. I am open to new ideas and cooperation and contributions. Lets see what happens. Thank you On Mon, Oct 18, 2021, 01:39 Matt Mahoney <mattmahone...@gmail.com> wrote: > The presentation showed some experimental results but I didn't get a > clear understanding of what experiments you did. How would your two > AGI brains perform on a text prediction problem (enwik9) or matching > labels to images (ImageNet), or some other well known benchmark? > > On Sun, Oct 17, 2021 at 8:54 AM Mohammadreza Alidoust > <class.alido...@gmail.com> wrote: > > > > Thanks Robert for your time and consideration. > > > > Do you mean the work lacks references? That statement was adopted from a > paper by Legg and Hutter. I mentioned their name. Could you please clarify? > > > > On Sun, Oct 17, 2021, 15:28 Quan Tesla <quantes...@gmail.com> wrote: > >> > >> Thank you for sharing your work. I'm not going to comment about the VI > and Brain I/II contents. However, your assertions/assumptions about what > comprises artificial general intelligence (the machine must perform a wide > range of general human-level tasks in diverse environments) lacks academic > specifity. > >> > >> There are numerous academic and contemporary sources you could have > accessed for constructing a measuring/research baseline to extrapolate for > AGI as it is today (checkpoint), but you chose not to do so in your > submission. > >> > >> Ultimately, your assertion of what measures for AGI possibly holds for > any system, even so, provided you actually specified the metrics as > differential values. > >> > >> I hope you find this feedback useful. > >> > >> Best regards > >> > >> Robert > >> > >> On 17 Oct 2021 12:42, "Mohammadreza Alidoust" <class.alido...@gmail.com> > wrote: > >>> > >>> My dear AGI friends, > >>> > >>> > >>> This is the link to my presentation about my paper entitled "AGI Brain > II: The Upgraded Version with Increased Versatility Index" at AGI-21 on > YouTube. I hope you enjoy it. > >>> > >>> https://youtu.be/GN14GIwS3Oc > >>> > >>> The Q&A session will be held on Monday 18th October (4th day of > AGI-21) at 8-9:30 AM or 4:30-6 PM PDT on YouTube or Zoom. > >>> > >>> I would appreciate your comments and questions about my work and I > will be happy to answer the questions. > >>> > >>> P.S.: the link to conference day 3 and 4 as well as the Q&A platform > will be announced by the AGI Society. > >>> > >>> > >>> Best regards, > >>> Mohammadreza Alidoust > > > > Artificial General Intelligence List / AGI / see discussions + > participants + delivery options Permalink > > -- > -- Matt Mahoney, mattmahone...@gmail.com ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T435e01e3365e303f-M2261917c2c3d16a45937c4a0 Delivery options: https://agi.topicbox.com/groups/agi/subscription
