HAMMING MASOFASI VA XOR MANTIQIY AMALI ASOSIDA FEW-SHOT LEARNING KLASSIFIKATSIYASINI AMALGA OSHIRISH
Mamagapurova S. S.
Annotatsiya
Zamonaviy Few-Shot Learning (FSL) modellari odatda haqiqiy sonlar fazosida (R^n)murakkab metrik hisob-kitoblarni amalga oshiradi, bu esa yuqori hisoblash resurslari va vaqt talab qiladi. Mazkur maqolada FSL masalalarini oʻta energiya tejamkor va tezkor hal qilish uchun ob’ektlar belgilarini binar (bit) vektorlar fazosiga oʻtkazish hamda klassifikatsiya jarayonini bit darajasidagi XOR (⊕) amali va Hamming masofasiga asoslash modeli taklif etilgan. Taklif etilayotgan yondashuv har bir sinf uchun bir nechta namunadan iborat bitli "prototip" (shablon) shakllantirish va qidiruv ob’ektini protsessor darajasida mantiqiy solishtirish tamoyiliga tayanadi. Eksperimental tahlillar modelning an’anaviy haqiqiy qiymatli metrikalarga qaraganda hisoblash vaqtini sezilarli darajada qisqartirishini va cheklangan namunalar sharoitida yuqori aniqlik (Accuracy) hamda barqarorlikni namoyish etishini koʻrsatdi.
Adabiyotlar
- Vinyals, O., Blundell, C., Lillicrap, T., & Wierstra, D. (2016). Matching networks for one shot learning. In Advances in Neural Information Processing Systems (NeurIPS), pp. 3630–3638.
- Snell, J., Swersky, K., & Zemel, R. (2017). Prototypical networks for few-shot learning. In Advances in Neural Information Processing Systems (NeurIPS), pp. 4077–4087.
- Hamming, R. W. (1950). Error detecting and error correcting codes. Bell System Technical Journal, 29(2), 147–160.
- González, C. G., Bonventi Jr, W., & Rodrigues, A. L. V. (2008). Density of Closed Balls in Real-Valued and Autometrized Boolean Spaces for Clustering Applications. LNAI, vol 5249. Springer, pp. 8–22.
- Kamilov, M. M., Fazylov, Sh. Kh., Mirzaev, N. M., & Radjabov, S. S. (2020). Models of recognition algorithms based on the estimation of feature interconnectedness. Tashkent.