We empirically characterize the performance of discriminative and generative LSTM models for text classification. We find that although RNNbased generative models are more powerful than their bag-of-words ancestors (e.g., they account for conditional dependencies across words in a document), they have higher asymptotic error rates than discriminatively trained RNN models. However we also find t...

We present a general framework for dependency parsing of Italian sentences based on a combination of discriminative and generative models. We use a state-of-the-art discriminative model to obtain a k -best list of candidate structures for the test sentences, and use the generative model to compute the probability of each candidate, and select the most probable one. We present the details of the...

In this paper, we propose a pattern language for Generative Participants. Recent education more and more requires methods to cultivate creativity in the students. A Generative Participant proposed in this paper is a teaching style aimed to cultivate creativity in its students. In a project-based environment, a Generative Participant teaches by actually participating in creative activities to st...

January is is the rst progress report for Project /// (Generative Kernels and Score Spaces for Classiication of Speech) within the Global Uncertainties Programme. is project combines the current generative models developed in the speech community with discriminative classiiers. An important aspect of the approach is that the generative models are used to deene a score-space that can be used as ...

In current agent systems agent migration is only possible between homogeneous systems supporting identical agent platforms, limiting an agent’s possibilities considerably. This paper revisits the notion of generative migration. Generative migration entails migration of an agent blueprint, instead of complete code. This approach relies on homogeneity of libraries on different platforms to re-inc...

Modular robots offer an important benefit in evolutionary robotics, which is to quickly evaluate evolved morphologies and control systems in reality. However, artificial evolution of simulated modular robotics is a difficult and time consuming task requiring significant computational power. While artificial evolution in virtual creatures has made use of powerful generative encodings, here we in...

Generative models for deep learning are promising both to improve understanding of the model, and yield training methods requiring fewer labeled samples. Recent works use generative model approaches to produce the deep net’s input given the value of a hidden layer several levels above. However, there is no accompanying “proof of correctness” for the generative model, showing that the feedforwar...

Density-aware generative algorithms learning from positive examples have verified high recall for visual object detection, but such generative methods suffer from excessive false positives which leads to low precision. Inspired by the recent success of detection-recognition pipeline with deep neural networks, this paper proposes a two-step framework by training a generative detector with positi...

The newly formed generative cell of the pollen grain of the African blood lily is spheroidal after its detachment from the pollen wall. Plastids are almost always excluded from the generative cell, while dictyosomes, mitochondria, ribosomes, endoplasmic reticulum, and microtubules are present throughout development of the cell. During the time that the generative cell is spheroidal, microtubule...

2 THE DISCRIMINATIVE MODEL SPACE CONTAINS THE GENERATIVE MODEL SPACE In the following we will show that the generative model with any set of parameters is equivalent to the discriminative model with an appropriately chosen set of parameters. In other words, the discriminative model with these particular parameters predicts the same context-specific substitution probabilities P (a|Ci) as the gen...