Deep Neural Networks (DNNs) have revolutionized many aspects of our lives. The use of DNNs is becoming ubiquitous including in softwares for image recognition, speech recognition, speech synthesis, language translation, to name a few. he training of DNN architectures however is computationally expensive. Once the model is created, its use in the intended application - the inference task, is computationally heavy too and the inference needs to be fast for real time use. For obtaining high performance today, the code of Deep Learning (DL) primitives optimized for specific architectures by expert programmers exposed via libraries is the norm. However, given the constant emergence of new DNN architectures, creating hand optimized code is expensive, slow and is not scalable. To address this performance-productivity challenge, in this paper we present compiler algorithms to automatically generate high performance implementations of DL primitives that closely match the performance of hand optimized libraries. We develop novel data reuse analysis algorithms using the polyhedral model to derive efficient execution schedules automatically. In addition, because most DL primitives use some variant of matrix multiplication at their core, we develop a flexible framework where it is possible to plug in library implementations of the same in lieu of a subset of the loops. We show that such a hybrid compiler plus a minimal library-use approach results in state-of-the-art performance. We develop compiler algorithms to also perform operator fusions that reduce data movement through the memory hierarchy of the computer system.

Next generation virtual assistants are envisioned to handle multimodal inputs (e.g., vision, memories of previous interactions, etc., in addition to the user's utterances), and perform multimodal actions (e.g., displaying a route in addition to generating the system's utterance). We introduce Situated Interactive MultiModal Conversations (SIMMC) as a new direction aimed at training agents that take multimodal actions grounded in a co-evolving multimodal input context in addition to the dialog history. We provide two SIMMC datasets totalling ~13K human-human dialogs (~169K utterances) using a multimodal Wizard-of-Oz (WoZ) setup, on two shopping domains: (a) furniture (grounded in a shared virtual environment) and, (b) fashion (grounded in an evolving set of images). We also provide logs of the items appearing in each scene, and contextual NLU and coreference annotations, using a novel and unified framework of SIMMC conversational acts for both user and assistant utterances. Finally, we present several tasks within SIMMC as objective evaluation protocols, such as Structural API Prediction and Response Generation. We benchmark a collection of existing models on these SIMMC tasks as strong baselines, and demonstrate rich multimodal conversational interactions. Our data, annotations, code, and models will be made publicly available.

Combinatorial optimization has found applications in numerous fields, from aerospace to transportation planning and economics. The goal is to find an optimal solution among a finite set of possibilities. The well-known challenge one faces with combinatorial optimization is the state-space explosion problem: the number of possibilities grows exponentially with the problem size, which makes solving intractable for large problems. In the last years, deep reinforcement learning (DRL) has shown its promise for designing good heuristics dedicated to solve NP-hard combinatorial optimization problems. However, current approaches have two shortcomings: (1) they mainly focus on the standard travelling salesman problem and they cannot be easily extended to other problems, and (2) they only provide an approximate solution with no systematic ways to improve it or to prove optimality. In another context, constraint programming (CP) is a generic tool to solve combinatorial optimization problems. Based on a complete search procedure, it will always find the optimal solution if we allow an execution time large enough. A critical design choice, that makes CP non-trivial to use in practice, is the branching decision, directing how the search space is explored. In this work, we propose a general and hybrid approach, based on DRL and CP, for solving combinatorial optimization problems. The core of our approach is based on a dynamic programming formulation, that acts as a bridge between both techniques. We experimentally show that our solver is efficient to solve two challenging problems: the traveling salesman problem with time windows, and the 4-moments portfolio optimization problem. Results obtained show that the framework introduced outperforms the stand-alone RL and CP solutions, while being competitive with industrial solvers.

After the discovery of adversarial examples and their adverse effects on deep learning models, many studies focused on finding more diverse methods to generate these carefully crafted samples. Although empirical results on the effectiveness of adversarial example generation methods against defense mechanisms are discussed in detail in the literature, an in-depth study of the theoretical properties and the perturbation effectiveness of these adversarial attacks has largely been lacking. In this paper, we investigate the objective functions of three popular methods for adversarial example generation: the L-BFGS attack, the Iterative Fast Gradient Sign attack, and Carlini & Wagner's attack (CW). Specifically, we perform a comparative and formal analysis of the loss functions underlying the aforementioned attacks while laying out large-scale experimental results on ImageNet dataset. This analysis exposes (1) the faster optimization speed as well as the constrained optimization space of the cross-entropy loss, (2) the detrimental effects of using the signature of the cross-entropy loss on optimization precision as well as optimization space, and (3) the slow optimization speed of the logit loss in the context of adversariality. Our experiments reveal that the Iterative Fast Gradient Sign attack, which is thought to be fast for generating adversarial examples, is the worst attack in terms of the number of iterations required to create adversarial examples in the setting of equal perturbation. Moreover, our experiments show that the underlying loss function of CW, which is criticized for being substantially slower than other adversarial attacks, is not that much slower than other loss functions. Finally, we analyze how well neural networks can identify adversarial perturbations generated by the attacks under consideration, hereby revisiting the idea of adversarial retraining on ImageNet.

Knowledge Graphs (KGs) have gained considerable attention recently from both academia and industry. In fact, incorporating graph technology and the copious of various graph datasets have led the research community to build sophisticated graph analytics tools. Therefore, the application of KGs has extended to tackle a plethora of real-life problems in dissimilar domains. Despite the abundance of the currently proliferated generic KGs, there is a vital need to construct domain-specific KGs. Further, quality and credibility should be assimilated in the process of constructing and augmenting KGs, particularly those propagated from mixed-quality resources such as social media data. This paper presents a novel credibility domain-based KG Embedding framework. This framework involves capturing a fusion of data obtained from heterogeneous resources into a formal KG representation depicted by a domain ontology. The proposed approach makes use of various knowledge-based repositories to enrich the semantics of the textual contents, thereby facilitating the interoperability of information. The proposed framework also embodies a credibility module to ensure data quality and trustworthiness. The constructed KG is then embedded in a low-dimension semantically-continuous space using several embedding techniques. The utility of the constructed KG and its embeddings is demonstrated and substantiated on link prediction, clustering, and visualisation tasks.

Learning low-dimensional latent state space dynamics models has been a powerful paradigm for enabling vision-based planning and learning for control. We introduce a latent dynamics learning framework that is uniquely designed to induce proportional controlability in the latent space, thus enabling the use of much simpler controllers than prior work. We show that our learned dynamics model enables proportional control from pixels, dramatically simplifies and accelerates behavioural cloning of vision-based controllers, and provides interpretable goal discovery when applied to imitation learning of switching controllers from demonstration.

Multitask learning is a methodology to boost generalization performance and also reduce computational intensity and memory usage. However, learning multiple tasks simultaneously can be more difficult than learning a single task because it can cause imbalance among tasks. To address the imbalance problem, we propose an algorithm to balance between tasks at the gradient level by applying gradient-based meta-learning to multitask learning. The proposed method trains shared layers and task-specific layers separately so that the two layers with different roles in a multitask network can be fitted to their own purposes. In particular, the shared layer that contains informative knowledge shared among tasks is trained by employing single gradient step update and inner/outer loop training to mitigate the imbalance problem at the gradient level. We apply the proposed method to various multitask computer vision problems and achieve state-of-the-art performance.

Autoregressive neural network models have been used successfully for sequence generation, feature extraction, and hypothesis scoring. This paper presents yet another use for these models: allocating more computation to more difficult inputs. In our model, an autoregressive model is used both to extract features and to predict observations in a stream of input observations. The surprisal of the input, measured as the negative log-likelihood of the current observation according to the autoregressive model, is used as a measure of input difficulty. This in turn determines whether a small, fast network, or a big, slow network, is used. Experiments on two speech recognition tasks show that our model can match the performance of a baseline in which the big network is always used with 15% fewer FLOPs.

Recent improvements in data collection volume from planetary and space physics missions have allowed the application of novel data science techniques. The Cassini mission for example collected over 600 gigabytes of scientific data from 2004 to 2017. This represents a surge of data on the Saturn system. Machine learning can help scientists work with data on this larger scale. Unlike many applications of machine learning, a primary use in planetary space physics applications is to infer behavior about the system itself. This raises three concerns: first, the performance of the machine learning model, second, the need for interpretable applications to answer scientific questions, and third, how characteristics of spacecraft data change these applications. In comparison to these concerns, uses of black box or un-interpretable machine learning methods tend toward evaluations of performance only either ignoring the underlying physical process or, less often, providing misleading explanations for it. We build off a previous effort applying a semi-supervised physics-based classification of plasma instabilities in Saturn's magnetosphere. We then use this previous effort in comparison to other machine learning classifiers with varying data size access, and physical information access. We show that incorporating knowledge of these orbiting spacecraft data characteristics improves the performance and interpretability of machine learning methods, which is essential for deriving scientific meaning. Building on these findings, we present a framework on incorporating physics knowledge into machine learning problems targeting semi-supervised classification for space physics data in planetary environments. These findings present a path forward for incorporating physical knowledge into space physics and planetary mission data analyses for scientific discovery.

Deep learning (DL) approaches are achieving extraordinary results in a wide range of domains but often require a massive collection of private data. Hence, methods for training neural networks on the joint data of different data owners, that keep each party's input confidential, are called for. We address the setting of horizontally distributed data in deep learning, where the participants' vulnerable intermediate results have to be processed in a privacy-preserving manner. The predominant scheme for this setting is based on homomorphic encryption (HE), and it is widely considered to be without alternative. In contrast to this, we demonstrate that a carefully chosen, less complex and computationally less expensive secure sum protocol in conjunction with default secure channels exhibits superior properties in terms of both collusion-resistance and runtime. Finally, we discuss several open research questions in the context of collaborative DL, which possibly might lead back to HE-based solutions.