La ibhodi lokuthuthukisa i-arduino Ivumela izinkulungwane nezinkulungwane zamaphrojekthi ahlukene ukuthi zenziwe, umkhawulo usemcabangweni womenzi ngamunye, nakuba futhi unokulinganiselwa ngokomzimba, njengenkumbulo, amandla okucubungula, njll. Kodwa-ke, kunemikhiqizo eminingi kanye namaphrojekthi okunweba amakhono abo ngokwengeziwe, njengoba kunjalo nge- kwethulwa okusha kwe-AIFES.
Manje, sibonga le phrojekthi edalwe ngu I-Fraunhofer IMS ye-Arduino, leli bhodi lomthombo ovulekile lizoba ne-a uhlaka lwe-artificial intelligence (AI) oluhlelwe ku-C, kusetshenziswa imitapo yolwazi ejwayelekile ye-GNU GCC. Abasebenzisi manje bazokwazi ukwengeza i-AIfES kuphrojekthi yabo ye-Arduino futhi bayihlanganise kusuka kumphathi womtapo wolwazi kusuka ku-IDE ukuze uyisebenzise ekuthuthukisweni kwakho, okuhlinzeka ngamathuba okusebenzisa ama-algorithms okufunda komshini ngisho nasezilawulini ezincane ezifana nebhodi Arduino UNO Ama-8-bit.
Lokhu kuzovumela abathuthukisi ukuthi bakhe inqwaba yamadivayisi e-IoT (I-inthanethi Yezinto) ezizimele kakhulu efwini futhi ezingahlakanipha kakhulu, futhi ngenhlonipho enkulu ngobumfihlo bakho, njengoba imisebenzi ingenziwa ungaxhunyiwe ku-inthanethi ebhodini le-Arduino ngaphandle kwesidingo. ukuthembela kumasevisi akude. Ngaphezu kwalokho, iphrojekthi ye-AIFES yethulwe ngaphansi Ilayisensi ye-GNU GPLv3, ngakho-ke imahhala, nakuba ivumela ilayisense ekhokhelwayo yamaphrojekthi okuhweba.
I-AifES ifana kakhulu futhi iyahambisana Izinhlaka ze-Python ML njengoba kunjalo nge-TensorFlow, Keras noma i-PyTorch, kodwa ukusebenza kwayo kuncishisiwe. Kodwa-ke, kule nguqulo ekhishiwe i-FNN (I-Feedforward Neural Networks) isivele isekelwe, ngaphezu kwalokho ivumela futhi ukuqaliswa kwemisebenzi ehlanganisiwe efana ne-ReLu, i-Sigmoid, noma i-Softmax. Ngakolunye uhlangothi, abathuthukisi basebenzela ukuletha esikhathini esizayo futhi ukuqaliswa kwe-ConvNet (Convolutional Neural Networks), okungenzeka kungathathi isikhathi eside kakhulu ukufika.
Ezinye nazo zifakiwe ama-algorithms wokuqeqesha evamile, njenge-SGD (Gradient Descent Optimizer) kanye ne-Adam Optimizer, phakathi kwabanye. Ngisho, nge-8-bit MCU, akukubi nhlobo ...