Meta Learning for Image Captioning
Authors: Nannan Li, Zhenzhong Chen, Shan Liu8626-8633
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on MS COCO validate the effectiveness of our approach when compared with the state-of-the-art methods. |
| Researcher Affiliation | Collaboration | Nannan Li,1 Zhenzhong Chen,1 Shan Liu2 1School of Remote Sensing and Information Engineering, Wuhan University, China 2Tencent Media Lab, Palo Alto, CA, USA |
| Pseudocode | No | The paper does not contain a clearly labeled 'Pseudocode' or 'Algorithm' section, nor does it present structured steps in a code-like format. |
| Open Source Code | No | The paper does not contain any explicit statement about releasing source code or a link to a code repository for the described methodology. |
| Open Datasets | Yes | We conduct experiments on MSCOCO (Lin et al. 2014) dataset with 123,287 labeled images. |
| Dataset Splits | Yes | We use public available splits (Karpathy and Fei-Fei 2015) which have 5000 randomly selected images for validation and test. |
| Hardware Specification | No | The paper does not provide any specific hardware details such as GPU or CPU models used for running the experiments. |
| Software Dependencies | No | The paper mentions optimizers and models (e.g., Adam optimizer, Faster R-CNN) but does not list specific software libraries or frameworks with their version numbers. |
| Experiment Setup | Yes | The number of hidden nodes of our network is set to 512 for the LSTM cell, with word embedding size of 512. We use Adam optimizer (Kingma and Ba 2014) with learning rate decay and set initial learning rate α = 0.01, β = 5e-4. λ is set to be 0.1... We use 0.5 dropout before the last layer and feed back 0.05 sampled words every 4 epochs starting from the the 10th epoch until reaching a 0.25 feeding back rate (Bengio et al. 2015). We add a batch normalization layer (Ioffe and Szegedy 2015) in the beginning of the LSTM model to accelerate training with mini-batch size of 50. |