Current massive datasets demand light-weight access for analysis. Discrete hashing methods are thus beneficial because they map high-dimensional data to compact binary codes that are efficient to store and process, while preserving semantic similarity. To optimize powerful deep learning methods for image hashing, gradient-based methods are required. Binary codes, however, are discrete and thus have no continuous derivatives. Relaxing the problem by solving it in a continuous space and then quantizing the solution is not guaranteed to yield separable binary codes. The quantization needs to be included in the optimization. In this paper we push for quantization: We optimize maximum class separability in the binary space. We introduce a margin on distances between dissimilar image pairs as measured in the binary space. In addition to pair-wise distances, we draw inspiration from Fisher's Linear Discriminant Analysis (Fisher LDA) to maximize the binary distances between classes and at the same time minimize the binary distance of images within the same class. Experiments on CIFAR-10, NUS-WIDE and ImageNet100 demonstrate compact codes comparing favorably to the current state of the art.

@en

Much of the observational data that we see around is, is ordered in space or time. For instance, video data, audio data or text data. This ordered data, called sequence data, calls for automatic analysis using supervised learning. Traditional single-observation supervised learning is challenged by sequence data, because (1) the length of sequence examples is often variable; (2) the sequence data may contain irrelevant segments which yield negative impact on the learning performance and (3) there exist temporal dependencies between consecutive observations in a sequence that need to be exploited by supervised learning on sequence data. This thesis introduces new models for supervised learning on sequence data that specifically address these challenges. We first propose a sequence classification model which is a graphical model using hidden variables to model the latent structure in the sequence data. It advances the state-of-the art by using the same number of hidden variables to model much more complex decision boundaries. Subsequently,we present a sequence classification model which is able to dealwith unsegmented sequences. The proposed model integrates ideas from attention models and gated recurrent neural networks. It is able to discern the salient segments and filter out the irrelevant ones, but it also measures the relevance of each time step of the sequence data to the final task. Finally, we propose an end-to-end model for age estimation from facial expression videos that performs feature learning and supervised learning for the final task jointly. Next we considered the supervised learning on paired sequences in which we want to predict whether the two sequences are similar. We combined ideas from sequence modeling and metric learning, and propose Siamese Recurrent Networks to learn a good similarity measure between two sequences. Our model is superior to current techniques that are based on handcrafted similarity measures or models using unsupervised learning. Finally, we present a model that predicts the preference of users for items in a recommendation system. In this case, two input sequences represent a pair of historic user and item data each with their own properties. The dependencies between the two sequences are modeled using an attention scheme.@en

We present a new model for multivariate time-series classification, called the hidden-unit logistic model (HULM), that uses binary stochastic hidden units to model latent structure in the data. The hidden units are connected in a chain structure that models temporal dependencies in the data. Compared with the prior models for time-series classification such as the hidden
conditional random field, our model can model very complex decision boundaries, because the number of latent states grows exponentially with the number of hidden units. We demonstrate the strong performance of our model in experiments on a variety of (computer vision) tasks, including handwritten character recognition, speech recognition, facial expression, and action recognition. We also present a state-of-the-art system for facial
action unit detection based on the HULM.@en

Capturing the temporal dynamics of user preferences over items is important for recommendation. Existing methods mainly assume that all time steps in user-item interaction history are equally relevant to recommendation, which however does not apply in real-world scenarios where user-item interactions can often happen accidentally. More importantly, they learn user and item dynamics separately, thus failing to capture their joint effects on user-item interactions. To better model user and item dynamics, we present the Interacting Attention-gated Recurrent Network (IARN) which adopts the attention model to measure the relevance of each time step. In particular, we propose a novel attention scheme to learn the attention scores of user and item history in an interacting way, thus to account for the dependencies between user and item dynamics in shaping user-item interactions. By doing so, IARN can selectively memorize different time steps of a user's history when predicting her preferences over different items. Our model can therefore provide meaningful interpretations for recommendation results, which could be further enhanced by auxiliary features. Extensive validation on real-world datasets shows that IARN consistently outperforms state-of-the-art methods.@en

Typical techniques for sequence classification are designed for well-segmented sequences which have been edited to remove noisy or irrelevant parts. Therefore, such methods cannot be easily applied on noisy sequences expected in real-world applications. In this paper, we present the Temporal Attention-Gated Model (TAGM) which integrates ideas from attention models and gated recurrent networks to better deal with noisy or unsegmented sequences. Specifically, we extend the concept of attention model to measure the relevance of each observation (time step) of a sequence. We then use a novel gated recurrent network to learn the hidden representation for the final prediction. An important advantage of our approach is interpretability since the temporal attention weights provide a meaningful value for the salience of each time step in the sequence. We demonstrate the merits of our TAGM approach, both for prediction accuracy and interpretability, on three different tasks: spoken digit recognition, text-based sentiment analysis and visual event recognition.@en