search

knowm / memristor-discovery

Java Application for automated memristor experiments and data collection based on Digilent's Analog Discovery 2 USB Oscilloscope and Knowm's Memristor Discovery Board.
Other
41 stars 10 forks source link

Getting an accuracy prediction for our own dataset #78

Open bakigkgz1 opened 4 months ago

bakigkgz1 commented 4 months ago

Hello. Can we run our own datasets on this system? For example, can we load the MNIST dataset into this system and get an accuracy prediction? Thank you very much in advance for your interest.

alex-nugent commented 4 months ago

Memristor Discovery V2 uses a chip with 16 discrete memristors to create a tiny neuron with 8 differential pair synapses. No, you cannot use that for accurate classification of MNIST.

timmolter commented 4 months ago

We did however do a more basic classify experiment, which you can read about here: https://github.com/knowm/memristor-discovery/blob/develop/src/main/resources/help/Classify12.md Also we have a video on it: https://vimeo.com/254209959.

bakigkgz1 commented 4 months ago

Dear Professor, hello. First of all, thank you very much for taking the time to answer my question. I watch your videos and I am also subscribed to your YouTube channel. As far as I understand, we cannot upload our own dataset and perform prediction tasks. However, in one of your videos, you imported a file with a DIMACS extension and conducted a trial with it, so I was wondering if we could work with different datasets for that. Thank you again, and I wish you good work.

Tim Molter @.***>, 16 May 2024 Per, 21:32 tarihinde şunu yazdı:

We did however do a more basic classify experiment, which you can read about here: https://github.com/knowm/memristor-discovery/blob/develop/src/main/resources/help/Classify12.md Also we have a video on it: https://vimeo.com/254209959.

— Reply to this email directly, view it on GitHub https://github.com/knowm/memristor-discovery/issues/78#issuecomment-2115940431, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6Q7AUCLGFSJPDL3ZZSBL3ZCT3U7AVCNFSM6AAAAABH2S426CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMJVHE2DANBTGE . You are receiving this because you authored the thread.Message ID: @.***>

-- Ögr.Gör.Baki GÖKGÖZ Gümüşhane Üniversitesi www.bakigokgoz.com @.***