We introduce a training method for both better word representation and performance, which we call GROVER (Gradual Rumination On the Vector with maskERs). The method is to gradually and iteratively add random noises to word embeddings while training a model. GROVER first starts from conventional training process, and then extracts the fine-tuned representations. Next, we gradually add random noises to the word representations and repeat the training process from scratch, but initialize with the noised word representations. Through the re-training process, we can mitigate some noises to be compensated and utilize other noises to learn better representations. As a result, we can get word representations further fine-tuned and specialized on the task. When we experiment with our method on 5 text classification datasets, our method improves model performances on most of the datasets. Moreover, we show that our method can be combined with other regularization techniques, further improving the model performance.

Bilevel optimizations are at the center of several important machine learning problems such as hyperparameter tuning, data denoising, few-shot learning, data poisoning. Different from simultaneous or multi-objective optimization, obtaining the exact descent direction for continuous bilevel optimization requires computing the inverse of the hessian of the lower-level cost function, even for first order methods. In this paper, we propose a new method for solving bilevel optimization, using the penalty function, which avoids computing the inverse of the hessian. We prove convergence of the method under mild conditions and show that it computes the exact hypergradient asymptotically. Small space and time complexity of our method allows us to solve large-scale bilevel optimization problems involving deep neural networks with up to 3.8M upper-level and 1.4M lower-level variables. We present results of our method for data denoising on MNIST/CIFAR10/SVHN datasets, for few-shot learning on Omniglot/Mini-Imagenet datasets and for training-data poisoning on MNIST/Imagenet datasets. In all experiments, our method outperforms or is comparable to previously proposed methods both in terms of accuracy and run-time.

Deep generative priors offer powerful models for complex-structured data, such as images, audio, and text. Using these priors in inverse problems typically requires estimating the input and/or hidden signals in a multi-layer deep neural network from observation of its output. While these approaches have been successful in practice, rigorous performance analysis is complicated by the non-convex nature of the underlying optimization problems. This paper presents a novel algorithm, Multi-Layer Vector Approximate Message Passing (ML-VAMP), for inference in multi-layer stochastic neural networks. ML-VAMP can be configured to compute maximum a priori (MAP) or approximate minimum mean-squared error (MMSE) estimates for these networks. We show that the performance of ML-VAMP can be exactly predicted in a certain high-dimensional random limit. Furthermore, under certain conditions, ML-VAMP yields estimates that achieve the minimum (i.e., Bayes-optimal) MSE as predicted by the replica method. In this way, ML-VAMP provides a computationally efficient method for multi-layer inference with an exact performance characterization and testable conditions for optimality in the large-system limit.

Learning generative models that span multiple data modalities, such as vision and language, is often motivated by the desire to learn more useful, generalisable representations that faithfully capture common underlying factors between the modalities. In this work, we characterise successful learning of such models as the fulfillment of four criteria: i) implicit latent decomposition into shared and private subspaces, ii) coherent joint generation over all modalities, iii) coherent cross-generation across individual modalities, and iv) improved model learning for individual modalities through multi-modal integration. Here, we propose a mixture-of-experts multimodal variational autoencoder (MMVAE) to learn generative models on different sets of modalities, including a challenging image-language dataset, and demonstrate its ability to satisfy all four criteria, both qualitatively and quantitatively.

--Thermal dynamics modeling has been a critical issue in building heating, ventilation, and air-conditioning (HV AC) systems, which can significantly affect the control and maintenance strategies. Due to the uniqueness of each specific building, traditional thermal dynamics modeling approaches heavily depending on physics knowledge cannot generalize well. This study proposes a deep supervised domain adaptation (DSDA) method for thermal dynamics modeling of building indoor temperature evolution and energy consumption. A long short term memory network based Sequence to Sequence scheme is pre-trained based on a large amount of data collected from a building and then adapted to another building which has a limited amount of data by applying the model fine-tuning. We use four publicly available datasets: SML and AHU for temperature evolution, long-term datasets from two different commercial buildings, termed as Building 1 and Building 2 for energy consumption. We show that the deep supervised domain adaptation is effective to adapt the pre-trained model from one building to another building and has better predictive performance than learning from scratch with only a limited amount of data.

The growing number of low-power smart devices in the Internet of Things is coupled with the concept of "Edge Computing", that is moving some of the intelligence, especially machine learning, towards the edge of the network. Enabling machine learning algorithms to run on resource-constrained hardware, typically on low-power smart devices, is challenging in terms of hardware (optimized and energy-efficient integrated circuits), algorithmic and firmware implementations. This paper presents FANN-on-MCU, an open-source toolkit built upon the Fast Artificial Neural Network (FANN) library to run lightweight and energy-efficient neural networks on microcontrollers based on both the ARM Cortex-M series and the novel RISC-V-based Parallel Ultra-Low-Power (PULP) platform. The toolkit takes multi-layer perceptrons trained with FANN and generates code targeted at execution on low-power microcontrollers either with a floating-point unit (i.e., ARM Cortex-M4F and M7F) or without (i.e., ARM Cortex M0-M3 or PULP-based processors). This paper also provides an architectural performance evaluation of neural networks on the most popular ARM Cortex-M family and the parallel RISC-V processor called Mr. Wolf. The evaluation includes experimental results for three different applications using a self-sustainable wearable multi-sensor bracelet. Experimental results show a measured latency in the order of only a few microseconds and a power consumption of few milliwatts while keeping the memory requirements below the limitations of the targeted microcontrollers. In particular, the parallel implementation on the octa-core RISC-V platform reaches a speedup of 22x and a 69% reduction in energy consumption with respect to a single-core implementation on Cortex-M4 for continuous real-time classification.

--Intrusion Detection System (IDS) is one of the most effective solutions for providing primary security services. IDSs are generally working based on attack signatures or by detecting anomalies. In this paper, we have presented AutoIDS, a novel yet efficient solution for IDS, based on a semi-supervised machine learning technique. AutoIDS can distinguish abnormal packet flows from normal ones by taking advantage of cascading two efficient detectors. These detectors are two encoder-decoder neural networks that are forced to provide a compressed and a sparse representation from the normal flows. In the test phase, failing these neural networks on providing compressed or sparse representation from an incoming packet flow, means such flow does not comply with the normal traffic and thus it is considered as an intrusion. For lowering the computational cost along with preserving the accuracy, a large number of flows are just processed by the first detector . In fact, the second detector is only used for difficult samples which the first detector is not confident about them. We have evaluated AutoIDS on the NSL-KDD benchmark as a widely-used and well-known dataset. The accuracy of AutoIDS is 90.17% showing its superiority compared to the other state-of-the-art methods. OW ADA YS, providing security services in different computer networks is an issue of paramount significance. The principal security services required by almost all of the communication networks, irrespective of their types, are confidentiality, authenticity, non-repudiation, integrity, and availability.

We study the invariance characteristics of pre-trained predictive models by empirically learning transformations on the input that leave the prediction function approximately unchanged. To learn invariance transformations, we minimize the Wasserstein distance between the predictive distribution conditioned on the data instances and the predictive distribution conditioned on the transformed data instances. To avoid finding degenerate or perturbative transformations, we further regularize by adding a similarity term between the data and its transformed values. Applying the proposed technique to clinical time series data, we discover variables that commonly-used LSTM models do not rely on for their prediction, especially when the LSTM is trained to be adversarially robust.

This is most apparent when auto-encoders are trained, where a network is trained to map the input data upon itself but is forced to project them into a lower-dimensional embedding space on the way (Vincent et al., 2010). In addition to the conventional fully connected layers, there are various special types of network connections. For example, in computer vision, convolu-tional layers are commonly used, which train multiple sliding windows that move over the image data and process just a part of the image at a time, thereby learning to recognize local features. These layers are subsequently abstracted into more and more complex visual patterns (Krizhevsky et al., 2017). For temporal data, one can use recurrent neural networks, which do not make predictions for individual input vectors, but for a sequence of input vectors. To do so, they allow feeding abstracted information from previous data points forward to the next layers.

This paper proposes a step toward obtaining general models of knowledge for facial analysis, by addressing the question of multi-source transfer learning. More precisely, the proposed approach consists in two successive training steps: the first one consists in applying a combination operator to define a common embedding for the multiple sources materialized by different existing trained models. The proposed operator relies on an auto-encoder, trained on a large dataset, efficient both in terms of compression ratio and transfer learning performance. In a second step we exploit a distillation approach to obtain a lightweight student model mimicking the collection of the fused existing models. This model outperforms its teacher on novel tasks, achieving results on par with state-of-the-art methods on 15 facial analysis tasks (and domains), at an affordable training cost. Moreover, this student has 75 times less parameters than the original teacher and can be applied to a variety of novel face-related tasks.