Whether you are an algorithm developer in a growing startup company, a data scientist in a university research lab, or a kaggle hobbyist, there may come a point in time when the training resources that you have onsite no longer meet your training demands. In this post we target development teams that are (finally) ready to move their machine learning (ML) workloads to the cloud. We will discuss some of the important decisions that need to made during this big transition. Naturally, any attempt to encompass all of the steps of such an endeavor is doomed to fail. Machine learning projects come in many shapes and forms and as their complexity increases so does the undertaking of making such a significant change as migrating to the cloud. In this post we will highlight what we believe to be some of the most important considerations that are common to most typical deep learning projects.

This article is part of our reviews of AI research papers, a series of posts that explore the latest findings in artificial intelligence. Contrastive learning (CL) is a machine learning technique that has gained popularity in the past few years because it reduces the need for annotated data, one of the main pain points of developing ML models. But due to its peculiarities, contrastive learning presents security challenges that are different from those found in supervised machine learning. Machine learning and security researchers are worried about the effect of adversarial attacks on ML models trained through contrastive learning. Accepted at NeurIPS 2021, the paper introduces a new technique that helps protect contrastive learning models against adversarial attacks while also preserving their accuracy. Supervised learning, the traditional way of training ML models, requires large sets of labeled data.

Self-supervised learning algorithms, including BERT and SimCLR, have enabled significant strides in fields like natural language processing, computer vision, and speech processing. However, these algorithms are domain-specific, meaning that new self-supervised learning algorithms must be developed for each new setting, including myriad healthcare, scientific, and multimodal domains. To catalyze progress toward domain-agnostic methods, we introduce DABS: a Domain-Agnostic Benchmark for Self-supervised learning. To perform well on DABS, an algorithm is evaluated on seven diverse domains: natural images, multichannel sensor data, English text, speech recordings, multilingual text, chest x-rays, and images with text descriptions. Each domain contains an unlabeled dataset for pretraining; the model is then is scored based on its downstream performance on a set of labeled tasks in the domain. We also present e-Mix and ShED: two baseline domain-agnostic algorithms; their relatively modest performance demonstrates that significant progress is needed before self-supervised learning is an out-of-the-box solution for arbitrary domains. Code for benchmark datasets and baseline algorithms is available at https://github.com/alextamkin/dabs.

We propose a novel framework to classify large-scale time series data with long duration. Long time seriesclassification (L-TSC) is a challenging problem because the dataoften contains a large amount of irrelevant information to theclassification target. The irrelevant period degrades the classifica-tion performance while the relevance is unknown to the system.This paper proposes an uncertainty-aware multiple instancelearning (MIL) framework to identify the most relevant periodautomatically. The predictive uncertainty enables designing anattention mechanism that forces the MIL model to learn from thepossibly discriminant period. Moreover, the predicted uncertaintyyields a principled estimator to identify whether a prediction istrustworthy or not. We further incorporate another modality toaccommodate unreliable predictions by training a separate modelbased on its availability and conduct uncertainty aware fusion toproduce the final prediction. Systematic evaluation is conductedon the Automatic Identification System (AIS) data, which is col-lected to identify and track real-world vessels. Empirical resultsdemonstrate that the proposed method can effectively detect thetypes of vessels based on the trajectory and the uncertainty-awarefusion with other available data modality (Synthetic-ApertureRadar or SAR imagery is used in our experiments) can furtherimprove the detection accuracy.

We present the design of a learning-based compliance controller for assembly operations for industrial robots. We propose a solution within the general setting of learning from demonstration (LfD), where a nominal trajectory is provided through demonstration by an expert teacher. This can be used to learn a suitable representation of the skill that can be generalized to novel positions of one of the parts involved in the assembly, for example the hole in a peg-in-hole (PiH) insertion task. Under the expectation that this novel position might not be entirely accurately estimated by a vision or other sensing system, the robot will need to further modify the generated trajectory in response to force readings measured by means of a force-torque (F/T) sensor mounted at the wrist of the robot or another suitable location. Under the assumption of constant velocity of traversing the reference trajectory during assembly, we propose a novel accommodation force controller that allows the robot to safely explore different contact configurations. The data collected using this controller is used to train a Gaussian process model to predict the misalignment in the position of the peg with respect to the target hole. We show that the proposed learning-based approach can correct various contact configurations caused by misalignment between the assembled parts in a PiH task, achieving high success rate during insertion. We show results using an industrial manipulator arm, and demonstrate that the proposed method can perform adaptive insertion using force feedback from the trained machine learning models.

Neuro-Symbolic AI, which is alternatively called composite AI, is a relatively new term for a well-established concept with enormous significance for almost any enterprise application of Artificial Intelligence. By combining AI's statistical foundation (exemplified by machine learning) with its knowledge foundation (exemplified by knowledge graphs and rules), organizations get the most effective cognitive analytics results with the least amount of headaches--and cost. Pairing these two historical pillars of AI is essential to maximizing investments in these technologies and in data themselves. By itself, rules-based symbolic reasoning doesn't improve over time. Together, these AI approaches create total machine intelligence with logic-based systems that get better with each application.

The common approach to compositional data analysis is to transform the data by means of logratios. Logratios between pairs of compositional parts (pairwise logratios) are the easiest to interpret in many research problems. When the number of parts is large, some form of logratio selection is a must, for instance by means of an unsupervised learning method based on a stepwise selection of the pairwise logratios that explain the largest percentage of the logratio variance in the compositional dataset. In this article we present three alternative stepwise supervised learning methods to select the pairwise logratios that best explain a dependent variable in a generalized linear model, each geared for a specific problem. The first method features unrestricted search, where any pairwise logratio can be selected. This method has a complex interpretation if some pairs of parts in the logratios overlap, but it leads to the most accurate predictions. The second method restricts parts to occur only once, which makes the corresponding logratios intuitively interpretable. The third method uses additive logratios, so that $K-1$ selected logratios involve exactly $K$ parts. This method in fact searches for the subcomposition with the highest explanatory power. Once the subcomposition is identified, the researcher's favourite logratio representation may be used in subsequent analyses, not only pairwise logratios. Our methodology allows logratios or non-compositional covariates to be forced into the models based on theoretical knowledge, and various stopping criteria are available based on information measures or statistical significance with the Bonferroni correction. We present an illustration of the three approaches on a dataset from a study predicting Crohn's disease. The first method excels in terms of predictive power, and the other two in interpretability.

As \emph{artificial intelligence} (AI) systems are increasingly involved in decisions affecting our lives, ensuring that automated decision-making is fair and ethical has become a top priority. Intuitively, we feel that akin to human decisions, judgments of artificial agents should necessarily be grounded in some moral principles. Yet a decision-maker (whether human or artificial) can only make truly ethical (based on any ethical theory) and fair (according to any notion of fairness) decisions if full information on all the relevant factors on which the decision is based are available at the time of decision-making. This raises two problems: (1) In settings, where we rely on AI systems that are using classifiers obtained with supervised learning, some induction/generalization is present and some relevant attributes may not be present even during learning. (2) Modeling such decisions as games reveals that any -- however ethical -- pure strategy is inevitably susceptible to exploitation. Moreover, in many games, a Nash Equilibrium can only be obtained by using mixed strategies, i.e., to achieve mathematically optimal outcomes, decisions must be randomized. In this paper, we argue that in supervised learning settings, there exist random classifiers that perform at least as well as deterministic classifiers, and may hence be the optimal choice in many circumstances. We support our theoretical results with an empirical study indicating a positive societal attitude towards randomized artificial decision-makers, and discuss some policy and implementation issues related to the use of random classifiers that relate to and are relevant for current AI policy and standardization initiatives.

Reconstructing medical images from partial measurements is an important inverse problem in Computed Tomography (CT) and Magnetic Resonance Imaging (MRI). Existing solutions based on machine learning typically train a model to directly map measurements to medical images, leveraging a training dataset of paired images and measurements. These measurements are typically synthesized from images using a fixed physical model of the measurement process, which hinders the generalization capability of models to unknown measurement processes. To address this issue, we propose a fully unsupervised technique for inverse problem solving, leveraging the recently introduced score-based generative models. Specifically, we first train a score-based generative model on medical images to capture their prior distribution. Given measurements and a physical model of the measurement process at test time, we introduce a sampling method to reconstruct an image consistent with both the prior and the observed measurements. Our method does not assume a fixed measurement process during training, and can thus be flexibly adapted to different measurement processes at test time. Empirically, we observe comparable or better performance to supervised learning techniques in several medical imaging tasks in CT and MRI, while demonstrating significantly better generalization to unknown measurement processes. Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) are commonly used imaging tools for medical diagnosis. Reconstructing CT and MRI images from raw measurements (sinograms for CT and k-spaces for MRI) are well-known inverse problems. Specifically, measurements in CT are given by X-ray projections of an object from various directions, and measurements in MRI are obtained by inspecting the Fourier spectrum of an object with magnetic fields.

Supervised learning can learn large representational spaces, which are crucial for handling difficult learning tasks. However, due to the design of the model, classical image classification approaches struggle to generalize to new problems and new situations when dealing with small datasets. In fact, supervised learning can lose the location of image features which leads to supervision collapse in very deep architectures. In this paper, we investigate how self-supervision with strong and sufficient augmentation of unlabeled data can train effectively the first layers of a neural network even better than supervised learning, with no need for millions of labeled data. The main goal is to disconnect pixel data from annotation by getting generic task-agnostic low-level features. Furthermore, we look into Vision Transformers (ViT) and show that the low-level features derived from a self-supervised architecture can improve the robustness and the overall performance of this emergent architecture. We evaluated our method on one of the smallest open-source datasets STL-10 and we obtained a significant boost of performance from 41.66% to 83.25% when inputting low-level features from a self-supervised learning architecture to the ViT instead of the raw images.