In this paper, we propose a object detection method to solve the grasp planning problem when soft and rigid objects are mixed together. The object detection method is the extension of YOLOv3, which can detect objects covered by towels via rotated bo…

In this section, we report our results on computational metrics, hyperparameters and training configurations for our models. Table 1 shows the results of the perplexity score evaluation of the evaluated models, Table 2 shows hyperparameters for each…

In this work we discuss how a small, targeted dataset and a general seq2seq architecture can be used to build dialogue systems with human-like attributes. The system was assessed by human raters with high scores for personality, sarcasm and humor, h…

In this work, we presented the Hypermap framework, that allows to manage multiple layers of different types of maps. We furthermore introduced a mapping algorithm that creates polygonal semantic maps from RGB-D camera images. The semantic mapping pr…

In this paper, we propose three effective techniques for transfer learning using pivot-based parallel data. The principle is to pre-train NMT models with source-pivot and pivot-target parallel data and transfer the source encoder and the target deco…

While most work in the field of neuro-evolution focuses on the evolution of neural architectures that can potentially outperform hand-crafted designs, this work showcases the potential of neuro-evolution for a different use: a robust analysis and in…

This paper has presented an algorithm that consists in a generic optimisation of a firing graph, in an attempt to solve the abstract task of identifying latent factor’s activations. Furthermore it has provided theoretical certitude on the effectivne…

We have proposed a margin-based triplet embedding regularization (MTER) method to improve the robustness of DNNs. Experiments on MNIST [16], CASIA-WebFace [54], VGGFace2 [4] and MS1M [15] have demonstrated the effectiveness of our method in simple o…

While our approach yields improved results for most documents, there are some limitations that may affect the quality in some scenarios. First, we do not consider the document boundary during processing, therefore if the image is left uncropped, the…

The research of SM was partially supported by European Research Council Consolidator grant ERCCoG 770880: COMANFLO.

We hope that CodeSearchNet is a good step towards engaging with the machine learning, IR and NLP communities towards developing new machine learning models that understand source code and natural language. Despite the fact this report gives emphasis…

We proposed a powerful and efficient algorithm termed Genetic Neural Architecture Search (GeNAS) for sperm abnormality detection. Besides, a novel fitness (objective) function named GeNAS weighting factor (GeNAS-WF) introduced to evaluate the approp…

We thank Adhi Kuncoro for his helpful suggestions on this writeup.

In this paper, we propose Nonlinear Rank Approximation loss (NRA) for deep metric learning, which significantly improves upon existing approaches like Triplet, Lifted Structured, and N-pair loss. Our approach uses all batch elements as anchor points…