Considering the abundance of unlabeled speech data and the high labeling costs, unsupervised learning methods can be essential for better system development. One of the most successful methods is contrastive self-supervised methods, which require negative sampling: sampling alternative samples to contrast with the current sample (anchor). However, it is hard to ensure if all the negative samples belong to classes different from the anchor class without labels. This paper applies a non-contrastive self-supervised learning method on an unlabeled speech corpus to learn utterance-level embeddings. We used DIstillation with NO labels (DINO), proposed in computer vision, and adapted it to the speech domain. Unlike the contrastive methods, DINO does not require negative sampling. These embeddings were evaluated on speaker verification and emotion recognition. In speaker verification, the unsupervised DINO embedding with cosine scoring provided 4.38% EER on the VoxCeleb1 test trial. This outperforms the best contrastive self-supervised method by 40% relative in EER. An iterative pseudo-labeling training pipeline, not requiring speaker labels, further improved the EER to 1.89%. In emotion recognition, the DINO embedding performed 60.87, 79.21, and 56.98% in micro-f1 score on IEMOCAP, Crema-D, and MSP-Podcast, respectively. The results imply the generality of the DINO embedding to different speech applications.

The identification and classification of transitions in topological and microstructural regimes in pattern-forming processes are critical for understanding and fabricating microstructurally precise novel materials in many application domains. Unfortunately, relevant microstructure transitions may depend on process parameters in subtle and complex ways that are not captured by the classic theory of phase transition. While supervised machine learning methods may be useful for identifying transition regimes, they need labels which require prior knowledge of order parameters or relevant structures describing these transitions. Motivated by the universality principle for dynamical systems, we instead use a self-supervised approach to solve the inverse problem of predicting process parameters from observed microstructures using neural networks. This approach does not require predefined, labeled data about the different classes of microstructural patterns or about the target task of predicting microstructure transitions. We show that the difficulty of performing the inverse-problem prediction task is related to the goal of discovering microstructure regimes, because qualitative changes in microstructural patterns correspond to changes in uncertainty predictions for our self-supervised problem. We demonstrate the value of our approach by automatically discovering transitions in microstructural regimes in two distinct pattern-forming processes: the spinodal decomposition of a two-phase mixture and the formation of concentration modulations of binary alloys during physical vapor deposition of thin films. This approach opens a promising path forward for discovering and understanding unseen or hard-to-discern transition regimes, and ultimately for controlling complex pattern-forming processes.

This work tackles Weakly Supervised Anomaly detection, in which a predictor is allowed to learn not only from normal examples but also from a few labeled anomalies made available during training. In particular, we deal with the localization of anomalous activities within the video stream: this is a very challenging scenario, as training examples come only with video-level annotations (and not frame-level). Several recent works have proposed various regularization terms to address it i.e. by enforcing sparsity and smoothness constraints over the weakly-learned frame-level anomaly scores. In this work, we get inspired by recent advances within the field of self-supervised learning and ask the model to yield the same scores for different augmentations of the same video sequence. We show that enforcing such an alignment improves the performance of the model on XD-Violence.

This work explores how self-supervised learning can be universally used to discover speaker-specific features towards enabling personalized speech enhancement models. We specifically address the few-shot learning scenario where access to cleaning recordings of a test-time speaker is limited to a few seconds, but noisy recordings of the speaker are abundant. We develop a simple contrastive learning procedure which treats the abundant noisy data as makeshift training targets through pairwise noise injection: the model is pretrained to maximize agreement between pairs of differently deformed identical utterances and to minimize agreement between pairs of similarly deformed nonidentical utterances. Our experiments compare the proposed pretraining approach with two baseline alternatives: speaker-agnostic fully-supervised pretraining, and speaker-specific self-supervised pretraining without contrastive loss terms. Of all three approaches, the proposed method using contrastive mixtures is found to be most robust to model compression (using 85% fewer parameters) and reduced clean speech (requiring only 3 seconds).

Progress in automated microscopy and quantitative image analysis has promoted high-content screening (HCS) as an efficient drug discovery and research tool. While HCS offers to quantify complex cellular phenotypes from images at high throughput, this process can be obstructed by image aberrations such as out-of-focus image blur, fluorophore saturation, debris, a high level of noise, unexpected auto-fluorescence or empty images. While this issue has received moderate attention in the literature, overlooking these artefacts can seriously hamper downstream image processing tasks and hinder detection of subtle phenotypes. It is therefore of primary concern, and a prerequisite, to use quality control in HCS. In this work, we evaluate deep learning options that do not require extensive image annotations to provide a straightforward and easy to use semi-supervised learning solution to this issue. Concretely, we compared the efficacy of recent self-supervised and transfer learning approaches to provide a base encoder to a high throughput artefact image detector. The results of this study suggest that transfer learning methods should be preferred for this task as they not only performed best here but present the advantage of not requiring sensitive hyperparameter settings nor extensive additional training.

Self-Supervised Learning (SSL) is a new paradigm for learning discriminative representations without labelled data and has reached comparable or even state-of-the-art results in comparison to supervised counterparts. Contrastive Learning (CL) is one of the most well-known approaches in SSL that attempts to learn general, informative representations of data. CL methods have been mostly developed for applications in computer vision and natural language processing where only a single sensor modality is used. A majority of pervasive computing applications, however, exploit data from a range of different sensor modalities. While existing CL methods are limited to learning from one or two data sources, we propose COCOA (Cross mOdality COntrastive leArning), a self-supervised model that employs a novel objective function to learn quality representations from multisensor data by computing the cross-correlation between different data modalities and minimizing the similarity between irrelevant instances. We evaluate the effectiveness of COCOA against eight recently introduced state-of-the-art self-supervised models, and two supervised baselines across five public datasets. We show that COCOA achieves superior classification performance to all other approaches. Also, COCOA is far more label-efficient than the other baselines including the fully supervised model using only one-tenth of available labelled data.

The role of chest X-ray (CXR) imaging, due to being more cost-effective, widely available, and having a faster acquisition time compared to CT, has evolved during the COVID-19 pandemic. To improve the diagnostic performance of CXR imaging a growing number of studies have investigated whether supervised deep learning methods can provide additional support. However, supervised methods rely on a large number of labeled radiology images, which is a time-consuming and complex procedure requiring expert clinician input. Due to the relative scarcity of COVID-19 patient data and the costly labeling process, self-supervised learning methods have gained momentum and has been proposed achieving comparable results to fully supervised learning approaches. In this work, we study the effectiveness of self-supervised learning in the context of diagnosing COVID-19 disease from CXR images. We propose a multi-feature Vision Transformer (ViT) guided architecture where we deploy a cross-attention mechanism to learn information from both original CXR images and corresponding enhanced local phase CXR images. We demonstrate the performance of the baseline self-supervised learning models can be further improved by leveraging the local phase-based enhanced CXR images. By using 10\% labeled CXR scans, the proposed model achieves 91.10\% and 96.21\% overall accuracy tested on total 35,483 CXR images of healthy (8,851), regular pneumonia (6,045), and COVID-19 (18,159) scans and shows significant improvement over state-of-the-art techniques. Code is available https://github.com/endiqq/Multi-Feature-ViT

Entailment trees have been proposed to simulate the human reasoning process of explanation generation in the context of open--domain textual question answering. However, in practice, manually constructing these explanation trees proves a laborious process that requires active human involvement. Given the complexity of capturing the line of reasoning from question to the answer or from claim to premises, the issue arises of how to assist the user in efficiently constructing multi--level entailment trees given a large set of available facts. In this paper, we frame the construction of entailment trees as a sequence of active premise selection steps, i.e., for each intermediate node in an explanation tree, the expert needs to annotate positive and negative examples of premise facts from a large candidate list. We then iteratively fine--tune pre--trained Transformer models with the resulting positive and tightly controlled negative samples and aim to balance the encoding of semantic relationships and explanatory entailment relationships. Experimental evaluation confirms the measurable efficiency gains of the proposed active fine--tuning method in facilitating entailment trees construction: up to 20\% improvement in explanatory premise selection when compared against several alternatives.

Large Vision & Language models pretrained on web-scale data provide representations that are invaluable for numerous V&L problems. However, it is unclear how they can be used for reasoning about user-specific visual concepts in unstructured language. This problem arises in multiple domains, from personalized image retrieval to personalized interaction with smart devices. We introduce a new learning setup called Personalized Vision & Language (PerVL) with two new benchmark datasets for retrieving and segmenting user-specific "personalized" concepts "in the wild". In PerVL, one should learn personalized concepts (1) independently of the downstream task (2) allowing a pretrained model to reason about them with free language, and (3) does not require personalized negative examples. We propose an architecture for solving PerVL that operates by extending the input vocabulary of a pretrained model with new word embeddings for the new personalized concepts. The model can then reason about them by simply using them in a sentence. We demonstrate that our approach learns personalized visual concepts from a few examples and can effectively apply them in image retrieval and semantic segmentation using rich textual queries.

Using the data from loop detector sensors for near-real-time detection of traffic incidents in highways is crucial to averting major traffic congestion. While recent supervised machine learning methods offer solutions to incident detection by leveraging human-labeled incident data, the false alarm rate is often too high to be used in practice. Specifically, the inconsistency in the human labeling of the incidents significantly affects the performance of supervised learning models. To that end, we focus on a data-centric approach to improve the accuracy and reduce the false alarm rate of traffic incident detection on highways. We develop a weak supervised learning workflow to generate high-quality training labels for the incident data without the ground truth labels, and we use those generated labels in the supervised learning setup for final detection. This approach comprises three stages. First, we introduce a data preprocessing and curation pipeline that processes traffic sensor data to generate high-quality training data through leveraging labeling functions, which can be domain knowledge-related or simple heuristic rules. Second, we evaluate the training data generated by weak supervision using three supervised learning models -- random forest, k-nearest neighbors, and a support vector machine ensemble -- and long short-term memory classifiers. The results show that the accuracy of all of the models improves significantly after using the training data generated by weak supervision. Third, we develop an online real-time incident detection approach that leverages the model ensemble and the uncertainty quantification while detecting incidents. Overall, we show that our proposed weak supervised learning workflow achieves a high incident detection rate (0.90) and low false alarm rate (0.08).