ÕªÒª£º 1. Sigmoid Êýѧ±í´ïʽ \[f(x) = \frac{1}{1+e^{-x}} \]ͼÏñ 2. Tanh Êýѧ±í´ïʽ \[f(x) = \frac{e^x - e^{-x}}{e^x + e^{-x}} \]ͼÏñ 3. ReLU Êýѧ±í´ïʽ \[f(x) = \max(0, x) \]ͼÏñ ÔĶÁÈ«ÎÄ
posted @ 2024-03-02 21:11 zh-jp ÔĶÁ(44) ÆÀÂÛ(0) ÍƼö(1) ±à¼­
ÕªÒª£º ʵÏÖ´úÂë import time import pydirectinput import keyboard if __name__ == '__main__': revolve = False while True: time.sleep(0.1) if keyboard.is_pressed(', ÔĶÁÈ«ÎÄ
posted @ 2024-02-24 10:12 zh-jp ÔĶÁ(263) ÆÀÂÛ(0) ÍƼö(1) ±à¼­
ÕªÒª£º Ç°ÑÔ ¹ØÓÚTransformerÔ­ÀíÓëÂÛÎĵĽéÉÜ£ºÏêϸÁ˽âTransformer£ºAttention Is All You Need PyTorchÖÐʵÏÖTransformerÄ£ÐÍ Ç°Ãæ½éÉÜÁË£¬Transformer Ä£ÐͽṹµÄʵÏÖ£¬ÕâÀï½éÉÜÏÂÂÛÎÄÖÐÌáµ½µÄѵÁ·²ßÂÔÓëÉèÖᣠÉèÖÃÎļþÃûΪtrainin ÔĶÁÈ«ÎÄ
posted @ 2024-02-07 15:43 zh-jp ÔĶÁ(142) ÆÀÂÛ(0) ÍƼö(0) ±à¼­
ÕªÒª£º ±¨´í ÔÚ¸´ÏÖ Transformer ´úÂëµÄѵÁ·½×¶Îʱ£¬·¢Éú±¨´í£º RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! ½â¾ö·½°¸ ͨ¹ý ÔĶÁÈ«ÎÄ
posted @ 2024-02-04 19:09 zh-jp ÔĶÁ(24) ÆÀÂÛ(0) ÍƼö(0) ±à¼­
ÕªÒª£º Ç°ÑÔ ¹ØÓÚTransformerÔ­ÀíÓëÂÛÎĵĽéÉÜ£ºÏêϸÁ˽âTransformer£ºAttention Is All You Need ¶ÔÓÚÂÛÎĸø³öµÄÄ£Ðͼܹ¹£¬Ê¹Óà PyTorch ·Ö±ðʵÏÖ¸÷¸ö²¿·Ö¡£ ÃüÃûtransformer.py£¬ÒýÈëµÄÏà¹Ø¿âº¯Êý£º import copy import torc ÔĶÁÈ«ÎÄ
posted @ 2024-02-01 16:34 zh-jp ÔĶÁ(538) ÆÀÂÛ(0) ÍƼö(2) ±à¼­
ÕªÒª£º Ç°ÑÔ Ê¹ÓÃopencv¶ÔͼÏñ½øÐвÙ×÷£¬ÒªÇ󣺣¨1£©¶¨Î»ÒøÐÐƱ¾ÝµÄËÄÌõ±ß£¬È»ºóÐýÕý¡££¨2£©¸ù¾Ý°æÃæ·ÖÎö£¬·Ö¸î³öСд½ð¶îÇøÓò¡£ ͼÏñУÕý Ê×ÏÈÊǶÔͼÏñµÄУÕý ¶ÁȡͼƬ ¶ÔͼƬ¶þÖµ»¯ ½øÐбßÔµ¼ì²â ¶Ô±ßÔµµÄ½øÐлô·òÂü±ä»» ½«±ä»»½á¹û´Ó¼«×ø±ê¿Õ¼äͶӰµ½µÑ¿¨¶û×ø±êµÃµ½Çãб½Ç ¸ù¾ÝÇãб½Ç¶ÔÖ÷ÌåУÕý import ÔĶÁÈ«ÎÄ
posted @ 2024-01-19 15:59 zh-jp ÔĶÁ(279) ÆÀÂÛ(0) ÍƼö(0) ±à¼­
ÕªÒª£º ÂÛÎÄ¡°Bi-directional Distribution Alignment for Transductive Zero-Shot Learning¡±Ìá³öBi-VAEGAN£¬ËüÒÔf-VAEGAN-D2ΪBaseline£¬½øÒ»²½·¢Õ¹ÁËTF-VAEGANͨ¹ýÀûÓÃËù¼ûÊý¾ÝºÍ·´À¡Ä£¿éÔöÇ¿Éú³ÉµÄÊÓ¾õÌØÕ÷˼ ÔĶÁÈ«ÎÄ
posted @ 2024-01-10 20:36 zh-jp ÔĶÁ(84) ÆÀÂÛ(0) ÍƼö(0) ±à¼­
ÕªÒª£º Ç°Ãæ½éÉÜÁ˽«VAE+GAN½â¾öÁãÑù±¾Ñ§Ï°µÄ·½·¨£ºf-VAEGAN-D2£¬ÕâÀï¼ÌÐøÌÖÂÛÒýÈëÉú³ÉÄ£ÐÍ´¦ÀíÁãÑù±¾Ñ§Ï°£¨Zero-shot Learning, ZSL£©ÎÊÌâ¡£ÂÛÎÄ¡°Latent Embedding Feedback and Discriminative Features for Zero-S ÔĶÁÈ«ÎÄ
posted @ 2024-01-10 16:44 zh-jp ÔĶÁ(121) ÆÀÂÛ(0) ÍƼö(0) ±à¼­
ÕªÒª£º ʹÓÃOpenCVʵÏÖÊÓƵȥ¶¶ ÕûÌå²½Ö裺 ÉèÖÃÊäÈëÊä³öÊÓƵ Ñ°ÕÒÖ¡Ö®¼äµÄÒƶ¯£ºÊ¹ÓÃopencvµÄÌØÕ÷¼ì²âÆ÷£¬¼ì²âÇ°Ò»Ö¡µÄÌØÕ÷£¬²¢Ê¹ÓÃLucas-Kanade¹âÁ÷Ëã·¨ÔÚÏÂÒ»Ö¡¸ú×ÙÕâЩÌØÕ÷£¬¸ù¾ÝÁ½×éµã£¬½«Ç°Ò»¸ö×ø±êϵӳÉäµ½µ±Ç°×ø±êϵÍê³É¸ÕÐÔ£¨Å·¼¸ÀïµÃ£©±ä»»£¬×îºóʹÓÃÊý×é¼Í¼֮֡¼äµÄÔ˶¯¡£ ¼ÆËãÖ¡Ö®¼äµÄƽ ÔĶÁÈ«ÎÄ
posted @ 2024-01-08 09:49 zh-jp ÔĶÁ(224) ÆÀÂÛ(0) ÍƼö(0) ±à¼­
ÕªÒª£º ËäÈ»f-VAEGAN-D2ÔÚÌâÄ¿ÖÐ˵¡°ÊÊÓÃÈa56爆大奖在线娱乐âÑù±¾¡±£¬µ«¶Ô±ÈµÄFew-shotÏà¹ØµÄʵÑé½ÏÉÙ£¬ÕâÀï½öÌÖÂÛÁãÑù±¾Ñ§Ï°µÄÇé¿ö¡£ 1. ±³¾°½éÉÜ ÓÉÓÚΪa56爆大奖在线娱乐¶ÔÏóÊÕ¼¯×ã¹»ÊýÁ¿µÄ¸ßÖÊÁ¿´ø±êÇ©Ñù±¾ÄÑÒÔʵÏÖ£¬Ê¹ÓÃÓÐÏ޵ıêÇ©½øÐÐѵÁ·Ñ§Ï°Ò»Ö±ÊÇÒ»¸öÖØÒªµÄÑо¿·½Ïò¡£ÁãÑù±¾Ñ§Ï°£¨Zero-Shot Learning, Z ÔĶÁÈ«ÎÄ
posted @ 2024-01-07 20:09 zh-jp ÔĶÁ(159) ÆÀÂÛ(0) ÍƼö(0) ±à¼­
ÕªÒª£º Ô­ÂÛÎÄÓÚ2023.11.6³·¸å£¬Ô­Òò£ºÈ±·¦ºÏ·¨µÄÊÚȨ£¬Ïê¼û´Ë´¦ Abstract ÔÚСÑù±¾Ñ§Ï°ÖУ¨Few-shot Learning, FSL£©ÖУ¬ÓÐͨ¹ýÀûÓöîÍâµÄÓïÒåÐÅÏ¢£¬ÈçÀàÃûµÄa56爆大奖在线娱乐Embedding£¬Í¨¹ý½«ÓïÒåÔ­ÐÍÓëÊÓ¾õÔ­ÐÍÏà½áºÏÀ´½â¾öÑù±¾Ï¡ÉÙµÄÎÊÌâ¡£µ«ÕâÖÖ·½·¨¿ÉÄÜ»áÓöµ½Ï¡ÓÐÑù±¾ÖÐѧµ½ÔëÉùÌØÕ÷ ÔĶÁÈ«ÎÄ
posted @ 2023-12-23 15:01 zh-jp ÔĶÁ(289) ÆÀÂÛ(0) ÍƼö(1) ±à¼­
ÕªÒª£º Éú³ÉÄ£ÐÍ ¸ø¶¨Êý¾Ý¼¯£¬Ï£ÍûÉú³ÉÄ£ÐͲúÉúÓëѵÁ·¼¯Í¬·Ö²¼µÄÐÂÑù±¾¡£¶ÔÓÚѵÁ·Êý¾Ý·þ´Ó\(p_{data}(x)\)£»¶ÔÓÚ²úÉúÑù±¾·þ´Ó\(p_{model}(x)\)¡£Ï£Íûѧµ½Ò»¸öÄ£ÐÍ\(p_{model}(x)\)Óë\(p_{data}(x)\)¾¡¿ÉÄܽӽü¡£ ÕâÒ²ÊÇÎ޼ලѧϰÖеÄÒ»¸öºËÐÄÎÊÌ⡪¡ªÃܶȹÀ¼ÆÎÊ ÔĶÁÈ«ÎÄ
posted @ 2023-12-10 21:49 zh-jp ÔĶÁ(222) ÆÀÂÛ(9) ÍƼö(1) ±à¼­
ÕªÒª£º Ç°ÑÔ a56爆大奖在线娱乐ʹÓÃPythonʵÏÖÁËPCAËã·¨£¬²¢Ê¹ÓÃORLÈËÁ³Êý¾Ý¼¯½øÐÐÁ˲âÊÔ²¢Êä³öÌØÕ÷Á³£¬¼òµ¥ÊµÏÖÁËÈËÁ³Ê¶±ðµÄ¹¦ÄÜ¡£ 1. ×¼±¸ ORLÈËÁ³Êý¾Ý¼¯¹²°üº¬40¸ö²»Í¬È˵Ä400ÕÅͼÏñ£¬ÊÇÔÚ1992Äê4ÔÂÖÁ1994Äê4ÔÂÆÚ¼äÓÉÓ¢¹ú½£ÇŵÄOlivettiÑо¿ÊµÑéÊÒ´´½¨¡£´ËÊý¾Ý¼¯°üº¬40¸öÀ࣬a56爆大奖在线娱乐Àຬ10ÕÅͼ ÔĶÁÈ«ÎÄ
posted @ 2023-11-30 20:43 zh-jp ÔĶÁ(171) ÆÀÂÛ(0) ÍƼö(0) ±à¼­
ÕªÒª£º ÂÛÎÄÁ´½Ó£ºMeta-Tuning Loss Functions and Data Augmentation for Few-Shot Object Detection Abstract Ïֽ׶εÄÉÙÑù±¾Ñ§Ï°¼¼Êõ¿ÉÒÔ·ÖΪÁ½Àࣺ»ùÓÚ΢µ÷£¨fine-tuning£©·½·¨ºÍ»ùÓÚԪѧϰ£¨meta-learning ÔĶÁÈ«ÎÄ
posted @ 2023-11-15 19:44 zh-jp ÔĶÁ(464) ÆÀÂÛ(0) ÍƼö(1) ±à¼­
ÕªÒª£º »ùÓÚԪѧϰ£¨Meta-Learning£©µÄ·½·¨£º Few-shotÎÊÌâ»ò³ÆΪFew-shotѧϰÊÇÏ£ÍûÄÜͨ¹ýÉÙÁ¿µÄ±ê×¢Êý¾ÝʵÏÖ¶ÔͼÏñµÄ·ÖÀ࣬ÊÇԪѧϰ(Meta-Learning)µÄa56爆大奖在线娱乐¡£ Few-shotѧϰ£¬²»ÊÇΪÁËѧϰ¡¢Ê¶±ðѵÁ·¼¯ÉϵÄÊý¾Ý£¬·º»¯µ½²âÊÔ¼¯£¬¶øÊÇΪÁËÈÃÄ£ÐÍѧ»áѧϰ¡£Ò²¾ÍÊÇÄ£ÐÍѵÁ·ºó£¬ ÔĶÁÈ«ÎÄ
posted @ 2023-11-11 11:01 zh-jp ÔĶÁ(1345) ÆÀÂÛ(0) ÍƼö(0) ±à¼­