技术控

    今日:0| 主题:63445
收藏本版 (1)
最新软件应用技术尽在掌握

[其他] AI generates abstract diagrams of IQ tests as good as 10th grade student

[复制链接]
Ange1浮夸 发表于 2016-10-1 05:59:01
371 11
Contextual RNN-GANs for Abstract Reasoning Diagram Generation

    Arnab Ghosh * , Viveka Kulharia * , Amitabha Mukerjee, Vinay Namboodiri, Mohit Bansal
    * Equal contribution
   Motivation

   
       
  • Understanding, predicting, and generating object motions and transformations is a core problem in artificial intelligence.
       
  • Modeling sequences of evolving images may provide better representations and models of motion and may ultimately be used for forecasting, simulation, or video generation.
       
  • Diagrammatic Abstract Reasoning is an avenue in which diagrams evolve in complex patterns and one needs to infer the underlying pattern sequence and generate the next image in the sequence.
       
   An Example with an Explanation

   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

   An explanation of the ground truth is that the dashed line first goes to the left, then to the right, and then on both sides, and also changes from single to double, hence the ground truth should have double dashed lines on both the sides. On the corners, the number of slanted lines increase by one after every two images, hence the ground truth should have four slant lines on both the corners.
   Some More Example Problems From DAT-DAR Dataset

   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

   The Model

   Contextual RNN-GAN

   
       
  • GANs have been shown to be useful in several image generation and manipulation tasks and hence it was a natural choice to prevent the model make fuzzy generations.
       
  • In Context-RNN-GAN, 'context' refers to the adversary receiving previous images (modeled as an RNN) and the generator is also an RNN. The name distinguishes it from our simpler RNN GAN model where the adversary is not contextual (as it only uses a single image) and only the generator is an RNN.
       
  • The discriminator is modeled as a GRU-RNN which gets all the preceding images to decide whether the generation by the Generator is the correct image for the timestep.
       
  • The generator is modeled as a GRU-RNN which tries to generate an image using the preceding images. It is guided by the contextual discriminator to produce real looking images.
       
   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

    The above figure corresponds to our Context-RNN-GAN model, where the generator G and the discriminator D (where D i represents its ith timestep snapshot) are both RNNs. G generates an image at every timestep while D receives all the preceding images as context to decide whether the current image output by G is real vs generated for that particular timestep. x i are the input images.
   Impact of Adversarial Loss

   
       
  • When using an L-2 loss function, some of the generated images were superimpositions of the component parts and were too cluttered.
       
  • When using an L-1 loss function, although it was sharper than using an L-2 loss, it was missing some components of the actual diagrams.
       
  • A weighted combination of L-1 and adversarial loss (as defined for the context based discriminator model described above) was used for the Context-RNN-GAN model to produce the best results based on empirical evaluation.
       
   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

   Some Generations (Contextual-RNN-GAN)

   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

   Modeling of Consecutive Timesteps using Siamese Networks for better accuracy

   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

   Comparison With Human Performance

                 College-grade     10th-grade             Age range     20-22     14-16             #Students     21     48             Mean     44.17%     36.67%             Std     16.67%     17.67%             Max     66.67%     75.00%             Min     8.33%     8.33%          Context-RNN-GAN with features obtained from Siamese CNN is competitive with humans in 10th grade in the sense that it is able to achieve accuracy of 35.4% when the generated features are compared with the features obtained from actual answer images. It needs to be noted that humans can see the options to get the best possible overall sequence of six images and hence can select the best choice while our model is just comparing the generations (obtained using sequence of five images in the problem) with options to get the best option. So, we can say that our model is very good generator and comparable to even 10th grade humans. An interesting aspect is that the model is never trained on the correct answers, it is just trained on multiple sequences from the problem images and still performs remarkably well.
   Interesting Cases

   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

   
       
  •   First GenerationIn the first example generation it is interesting to note that the model correctly predicted elements in off diagonal while faltered in the shape of the elements in leading diagonals.
       
  •   Second GenerationSecond example shows an interesting case whereby the image generated by our model is also plausible (if by symmetry it is considered that first and third the semicircular ring is solid and hence fourth and sixth should be solid) while the actual answer is of course plausible according to the reasoning that the (solid vs hollow) flipped in the first two cases then stayed the same for the next two timesteps. Even more interesting is its analysis of the spatial dynamics of the ball and the semicircular ring which it almost correctly captured.
       
  •   Third GenerationAnother very interesting case is the generation which it gets correct. However, in this case the answer figure is exactly similar to the second figure in the sequence. Therefore, it is not illustrative of the ability of the model to generate the sequence of the pattern.
       
   Application of the model to Moving-MNIST

   

AI generates abstract diagrams of IQ tests as good as 10th grade student

AI generates abstract diagrams of IQ tests as good as 10th grade student

    The model can be applied to video prediction tasks as illustrated by above figure. The Moving MNIST task consists of videos of two moving MNIST digits and the next frame has to be predicted from the preceding frames.
李明金 发表于 2016-10-1 07:25:16
楼下的,觉得比特币咋样啊?
回复 支持 反对

使用道具 举报

黄瑞琦 发表于 2016-10-1 17:40:16
那些路人甲乙丙丁在年生散场的剧场里将五彩纷呈和苍白无力潇洒的演绎。
回复 支持 反对

使用道具 举报

7548629 发表于 2016-10-1 17:59:54
能力就像瓜子仁,只有咬牙才能嗑出来。
回复 支持 反对

使用道具 举报

语嫣 发表于 2016-10-1 19:05:25
有空一起交流一下
回复 支持 反对

使用道具 举报

wzb1518 发表于 2016-10-2 06:56:34
秀起来~
回复 支持 反对

使用道具 举报

hwsfn 发表于 2016-10-3 02:46:34
楼下是射手
回复 支持 反对

使用道具 举报

anyword 发表于 2016-10-8 12:27:46
呵呵。。。
回复 支持 反对

使用道具 举报

回憶〤亂人心 发表于 2016-10-8 16:47:37
一语道尽爱情的残忍。情到深处人孤。疾苦使人成熟,强项的人会感悟爱的真谛,而懦弱的人徒生怅恨。
回复 支持 反对

使用道具 举报

一朵奇葩向阳开 发表于 2016-10-21 03:24:30
失而复得的东西。永远都是二手货。
回复 支持 反对

使用道具 举报

我要投稿

回页顶回复上一篇下一篇回列表
手机版/c.CoLaBug.com ( 粤ICP备05003221号 | 粤公网安备 44010402000842号 )

© 2001-2017 Comsenz Inc.

返回顶部 返回列表