Multi-Task Learning (MTL) is a learning paradigm in machine learning and its aim is to leverage useful information contained in multiple related tasks to help improve the generalization performance of all the tasks. In this paper, we give a survey for MTL. First, we classify different MTL algorithms into several categories, including feature learning approach, low-rank approach, task clustering approach, task relation learning approach, and decomposition approach, and then discuss the characteristics of each approach. In order to improve the performance of learning tasks further, MTL can be combined with other learning paradigms including semi-supervised learning, active learning, unsupervised learning, reinforcement learning, multi-view learning and graphical models. When the number of tasks is large or the data dimensionality is high, batch MTL models are difficult to handle this situation and online, parallel and distributed MTL models as well as dimensionality reduction and feature hashing are reviewed to reveal their computational and storage advantages. Many real-world applications use MTL to boost their performance and we review representative works. Finally, we present theoretical analyses and discuss several future directions for MTL.

After reviewing popular techniques used in supervised, unsupervised and semi-supervised machine learning, we focus on feature selection methods in these different contexts, especially the metrics used to assess the value of a feature or set of features, be it binary, continuous or categorical variables. We go in deeper details and review modern feature selection techniques for unsupervised learning, typically relying on entropy-like criteria. While these criteria are usually model-dependent or scale-dependent, we introduce a new model-free, data-driven methodology in this context, with an application to an interesting number theory problem (simulated data set) in which each feature has a known theoretical entropy. We also briefly discuss high precision computing as it is relevant to this peculiar data set, as well as units of information smaller than the bit.

Unsupervised learning may sound like a fancy way to say "let the kids learn on their own not to touch the hot oven" but it's actually a pattern-finding technique for mining inspiration from your data. It has nothing to do with machines running around without adult supervision, forming their own opinions about things. This post is beginner-friendly, but assumes you're familiar with the story so far: Check out the six instances above. These photographs are not accompanied by labels. No worries, your brain is pretty good at unsupervised learning.

Generative Adversarial Networks are powerful generative models that are able to model the manifold of natural images. We leverage this property to perform manifold regularization by approximating a variant of the Laplacian norm using a Monte Carlo approximation that is easily computed with the GAN. When incorporated into the semi-supervised feature-matching GAN we achieve state-of-the-art results for GAN-based semi-supervised learning on CIFAR-10 and SVHN benchmarks, with a method that is significantly easier to implement than competing methods. We also find that manifold regularization improves the quality of generated images, and is affected by the quality of the GAN used to approximate the regularizer.

Multimodal machine learning is a core research area spanning the language, visual and acoustic modalities. The central challenge in multimodal learning involves learning representations that can process and relate information from multiple modalities. In this paper, we propose two methods for unsupervised learning of joint multimodal representations using sequence to sequence (Seq2Seq) methods: a \textit{Seq2Seq Modality Translation Model} and a \textit{Hierarchical Seq2Seq Modality Translation Model}. We also explore multiple different variations on the multimodal inputs and outputs of these seq2seq models. Our experiments on multimodal sentiment analysis using the CMU-MOSI dataset indicate that our methods learn informative multimodal representations that outperform the baselines and achieve improved performance on multimodal sentiment analysis, specifically in the Bimodal case where our model is able to improve F1 Score by twelve points. We also discuss future directions for multimodal Seq2Seq methods.

Training a model to perform a task typically requires a large amount of data from the domains in which the task will be applied. However, it is often the case that data are abundant in some domains but scarce in others. Domain adaptation deals with the challenge of adapting a model trained from a data-rich source domain to perform well in a data-poor target domain. In general, this requires learning plausible mappings between domains. CycleGAN is a powerful framework that efficiently learns to map inputs from one domain to another using adversarial training and a cycle-consistency constraint. However, the conventional approach of enforcing cycle-consistency via reconstruction may be overly restrictive in cases where one or more domains have limited training data. In this paper, we propose an augmented cyclic adversarial learning model that enforces the cycle-consistency constraint through an external task specific model, which encourages the preservation of task-relevant content as opposed to exact reconstruction. This task specific model both relaxes the cycle-consistency constraint and complements the role of the discriminator during training, serving as an augmented information source for learning the mapping. In the experiment, we adopt a speech recognition model from each domain as the task specific model. Our approach improves absolute performance of speech recognition by $2\%$ for female speakers in the TIMIT dataset, where the majority of training samples are from male voices. We also explore digit classification with MNIST and SVHN in a low-resource setting and show that our approach improves absolute performance by $14\%$ and $4\%$ when adapting SVHN to MNIST and vice versa, respectively. Our approach also outperforms unsupervised domain adaptation methods, which require high-resource unlabeled target domain.

Generative adversarial models are powerful tools to model structure in complex distributions for a variety of tasks. Current techniques for learning generative models require an access to samples which have high quality, and advanced generative models are applied to generate samples from noisy training data through ambient modules. However, the modules are only practical for the output space of the generator, and their application in the hidden space is not well studied. In this paper, we extend the ambient module to the hidden space of the generator, and provide the uniqueness condition and the corresponding strategy for the ambient hidden generator in the adversarial training process. We report the practicality of the proposed method on the benchmark dataset.

Modern deep transfer learning approaches have mainly focused on learning generic feature vectors from one task that are transferable to other tasks, such as word embeddings in language and pretrained convolutional features in vision. However, these approaches usually transfer unary features and largely ignore more structured graphical representations. This work explores the possibility of learning generic latent relational graphs that capture dependencies between pairs of data units (e.g., words or pixels) from large-scale unlabeled data and transferring the graphs to downstream tasks. Our proposed transfer learning framework improves performance on various tasks including question answering, natural language inference, sentiment analysis, and image classification. We also show that the learned graphs are generic enough to be transferred to different embeddings on which the graphs have not been trained (including GloVe embeddings, ELMo embeddings, and task-specific RNN hidden unit), or embedding-free units such as image pixels.

With the proliferation of social media, fashion inspired from celebrities, reputed designers as well as fashion influencers has shortened the cycle of fashion design and manufacturing. However, with the explosion of fashion related content and large number of user generated fashion photos, it is an arduous task for fashion designers to wade through social media photos and create a digest of trending fashion. This necessitates deep parsing of fashion photos on social media to localize and classify multiple fashion items from a given fashion photo. While object detection competitions such as MSCOCO have thousands of samples for each of the object categories, it is quite difficult to get large labeled datasets for fast fashion items. Moreover, state-of-the-art object detectors do not have any functionality to ingest large amount of unlabeled data available on social media in order to fine tune object detectors with labeled datasets. In this work, we show application of a generic object detector, that can be pretrained in an unsupervised manner, on 24 categories from recently released Open Images V4 dataset. We first train the base architecture of the object detector using unsupervisd learning on 60K unlabeled photos from 24 categories gathered from social media, and then subsequently fine tune it on 8.2K labeled photos from Open Images V4 dataset. On 300 X 300 image inputs, we achieve 72.7% mAP on a test dataset of 2.4K photos while performing 11% to 17% better as compared to the state-of-the-art object detectors. We show that this improvement is due to our choice of architecture that lets us do unsupervised learning and that performs significantly better in identifying small objects.

We propose a new regularization method based on virtual adversarial loss: a new measure of local smoothness of the conditional label distribution given input. Virtual adversarial loss is defined as the robustness of the conditional label distribution around each input data point against local perturbation. Unlike adversarial training, our method defines the adversarial direction without label information and is hence applicable to semi-supervised learning. Because the directions in which we smooth the model are only "virtually" adversarial, we call our method virtual adversarial training (VAT). The computational cost of VAT is relatively low. For neural networks, the approximated gradient of virtual adversarial loss can be computed with no more than two pairs of forward- and back-propagations. In our experiments, we applied VAT to supervised and semi-supervised learning tasks on multiple benchmark datasets. With a simple enhancement of the algorithm based on the entropy minimization principle, our VAT achieves state-of-the-art performance for semi-supervised learning tasks on SVHN and CIFAR-10.