the archive

#benchmark evaluation

6 results

cs.SI2019

Assessing Sentiment of the Expressed Stance on Social Media

Abeer Aldayel, Walid Magdy

The paper investigates how sentiment polarity relates to stance expression on social media, showing that positive or negative sentiment does not reliably indicate supportive or opp…

#stance detection#sentiment analysis#social media#dataset creation

cs.CV2019

Generalised Zero-Shot Learning with a Classifier Ensemble over Multi-Modal Embedding Spaces

Rafael Felix, Ben Harwood, Michele Sasdelli +1

The paper proposes a generalized zero-shot learning approach that ensembles classifiers trained on visual, semantic, and joint embedding spaces, using a calibration step to combine…

#generalized zero-shot learning#classifier ensemble#multi-modal embeddings#model calibration

cs.LG2019

Sparse hierarchical representation learning on molecular graphs

Matthias Bal, Hagen Triendl, Mariana Assmann +6

The paper proposes new hierarchical pooling layers for graph neural networks that can incorporate edge features, and demonstrates their effectiveness on molecular graphs for drug d…

#graph neural networks#molecular representation#edge features#hierarchical pooling

cs.LG2019

Random Search and Reproducibility for Neural Architecture Search

Liam Li, Ameet Talwalkar

The paper introduces random search baselines—including early‑stopping and a weight‑sharing variant—for neural architecture search, demonstrates that they match or exceed state‑of‑t…

#neural architecture search#random search#weight sharing#reproducibility

cs.CL2019

A Hybrid Neural Network Model for Commonsense Reasoning

Pengcheng He, Xiaodong Liu, Weizhu Chen +1

The paper introduces a hybrid neural network that combines a masked language model and a semantic similarity model, sharing a BERT encoder, to improve performance on commonsense re…

#commonsense reasoning#hybrid neural networks#masked language model#semantic similarity

cs.CV2019

To Learn or Not to Learn: Analyzing the Role of Learning for Navigation in Virtual Environments

Noriyuki Kojima, Jia Deng

The paper compares learning‑based navigation agents with classical approaches in virtual environments, showing that classical methods generally outperform learned agents on standar…

#navigation#virtual environments#learning-based methods#classical methods