Sign In
Create Account
Back to top
 |
Chirurgia endoscopică a hipofizei
"Standardul de aur" în chirurgia hipofizară îl reprezintă endoscopia transnazală transsfenoidală. Echipa NeuroHope este antrenată în unul din cele mai mari centre de chirurgie a hipofizei din Europa, Spitalul Foch din Paris, centrul în care a fost introdus pentru prima dată endoscopul în chirurgia transnazală a hipofizei, de către neurochirurgul francez Guiot. Pe lângă tumorile cu origine hipofizară, prin tehnicile endoscopice transnazale pot fi abordate numeroase alte patologii neurochirurgicale. www.neurohope.ro |
Fizica Computationala
4 votes
Last Updated: Mar 28 2023 17:55, Started by
parabellum
, Sep 11 2016 11:15
·
7
7
#145
Posted 14 October 2022 - 22:38
|
Un alt algoritm cuantic adaugat, Bernstein-Vazirani: https://github.com/a...steinVazirani.h
Din criptografia cuantica, protocolul BB84: https://github.com/a...umCryptograpy.h |
#146
Posted 18 December 2022 - 11:52
|
Cred ca iau o pauza de la proiectul asta o vreme... as mai fi vrut sa adaug cateva chestii dar imi cam lipseste motivatia.
Chestiile care functioneaza, mentionate si in README: Quote
Grover's algorithm Deutsch-Jozsa algorithm Shor's algorithm Bernstein–Vazirani algorithm Quantum adder (half-adder & full-adder for 1-qubits and full adder for two N-qubit numbers) Quantum teleportation Superdense coding Quantum cryptography: BB84 protocol CHSH inequality violation Simulation of quantum simulation: Evolution in time with a Hamiltonian given as a sum of Pauli products with real coefficients Evolution of a 1D Gaussian packet in time, solving the 1D time-dependent Schrodinger equation using a Trotter decomposition and quantum Fourier transform |
#147
Posted 09 February 2023 - 14:18
|
Acum chiar ca am decis sa ma opresc o vreme din adaugarea de noi algoritmi
 Sunt sigur ca o sa revin si o sa mai adaug, dar deocamdata am in plan sa incep ceva pe machine learning, ca tot e la moda AIul zilele astea. Cam astea-s chestiile implementate pana acum: Quote
Grover's algorithm Deutsch-Jozsa algorithm Simon's algorithm Quantum Fourier Transform Shor's algorithm Bernstein–Vazirani algorithm Quantum teleportation Superdense coding Quantum cryptography: BB84 protocol CHSH inequality violation Quantum error correction: 3-qubit error correcting a qubit-flip 3-qubit error correcting a sign-flip Shor Code Quantum adders: Quantum half-adder for 1-qubit Quantum full-adder for 1-qubit Full adder for two N-qubit numbers Draper adder Draper adder with carry Simulation of quantum simulation: Evolution in time with a Hamiltonian given as a sum of Pauli products with real coefficients Evolution of a 1D Gaussian packet in time, solving the 1D time-dependent Schrodinger equation using a Trotter decomposition and quantum Fourier transform Paradoxes (although some of the above might be considered paradoxes as well): Quantum eraser General Elitzur-Vaidman Bomb tester/interaction free measurement/counterfactual computation Hardy's paradox |
#148
Posted 25 March 2023 - 18:24
|
Deja am adus proiectul de machine learning intr-o forma oarecum functionala, exista regresie liniara generalizata (cu exemple pe regresia liniara simpla, regresia liniara multivariata si multivariabila, regresie polinomiala, regresia logistica - inclusiv cu separatie neliniara, aka un cerc) si o retea neuronala simpla (multilayer perceptron). Exista teste pentru o retea simpla care invata xor (o problema 'clasica' pe tema asta) si o retea care invata sa separe clasele din cunoscutul Iris dataset. Cam asta afiseaza la iesire pe datasetul respectiv:
Quote
Neural Network test for the Iris dataset, Setosa is lineary separable from the other two, but the others two cannot be linearly separated Loss: 2.09869 Loss: 0.0511973 Loss: 0.0368485 Loss: 0.000125418 Loss: 0.00024591 Loss: 1.71955e-05 Loss: 1.42229e-05 Loss: 8.20373e-06 Loss: 7.25362e-06 Loss: 6.12088e-07 Loss: 6.93672e-07 Training set: Setosa true positives: 41, true negatives: 79, false positives: 0, false negatives: 0 Versicolor true positives: 38, true negatives: 82, false positives: 0, false negatives: 0 Virginica true positives: 41, true negatives: 79, false positives: 0, false negatives: 0 Setosa accuracy: 1 Versicolor accuracy: 1 Virginica accuracy: 1 Setosa specificity: 1 Versicolor specificity: 1 Virginica specificity: 1 Setosa precision: 1 Versicolor precision: 1 Virginica precision: 1 Setosa recall: 1 Versicolor recall: 1 Virginica recall: 1 Test set: Setosa true positives: 9, true negatives: 21, false positives: 0, false negatives: 0 Versicolor true positives: 11, true negatives: 18, false positives: 0, false negatives: 1 Virginica true positives: 8, true negatives: 20, false positives: 1, false negatives: 1 Setosa accuracy: 1 Versicolor accuracy: 0.966667 Virginica accuracy: 0.933333 Setosa specificity: 1 Versicolor specificity: 1 Virginica specificity: 0.952381 Setosa precision: 1 Versicolor precision: 1 Virginica precision: 0.888889 Setosa recall: 1 Versicolor recall: 0.916667 Virginica recall: 0.888889 Repository aici: https://github.com/a...MachineLearning Attached Files
|
#149
Posted 28 March 2023 - 17:37
|
Din ce in ce mai bune sanse pentru cod sa fie corect... am adaugat EMNIST dataset (o gramada de imagini cu litere si cifre scrise de mana, din care folosesc seturile cu cifre pentru teste) si cu softmax (simplu, nu retea neuronala) obtin acuratete > 94%, iar cu retea neuronala pot trece binisor de 98%.
Un singur nivel ascuns cu 32 de neuroni trece de 97%, doua nivele ascunse cu 1000 si respectiv 100 de neuroni deja dau rezultate mai bune decat ma asteptam (~98.75%). Asta fara sa insist pe tot felul de teste cu diversi parametri, initializari etc. Poate o sa adaug si dropout si nivele de normalizare... si poate chiar convolutii... dar asta e in plan pe termen mai lung. Edited by parabellum, 28 March 2023 - 17:46. |
#150
Posted 28 March 2023 - 17:55
|
Info aditionala despre reteaua neuronala antrenata pe EMNIST: In acest caz nivelele ascunse au activare 'leaky relu', nivelul de iesire e softmax, functia de cost e 'cross entropy' iar optimizarea se face cu 'minibatch stochastic gradient descent', mai precis, adam. Chestiile astea se pot ajusta, exista multe variante disponibile in proiect.
Edited by parabellum, 28 March 2023 - 17:55. |
Anunturi
▶ 0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users
-
- Change Theme
- English
- Mark Community Read
- Termene & conditii
- Confidentialitate
- Consimtamant
- Cookies
- Help
© 2001-2023 Softpedia. All rights reserved. Softpedia® and the Softpedia logo are registered trademarks of SoftNews Net SRL.
Ora implicita a Forumului Softpedia este ora standard a Romaniei (GMT+2). Mai multe informatii →
⚠ Postarile afisate pe Forumul Softpedia reprezinta o opinie subiectiva a membrilor care le-au publicat si nu a SoftNews Net SRL.