The field of visual representation learning has seen explosive growth in the past years, but its benefits in robotics have been surprisingly limited so far. Prior work uses generic visual representations as a basis to learn (task-specific) robot action policies (e.g. via behavior cloning). While the visual representations do accelerate learning, they are primarily used to encode visual observations. Thus, action information has to be derived purely from robot data, which is expensive to collect! In this work, we present a scalable alternative where the visual representations can help directly infer robot actions. We observe that vision encoders express relationships between image observations as distances (e.g. via embedding dot product) that could be used to efficiently plan robot behavior. We operationalize this insight and develop a simple algorithm for acquiring a distance function and dynamics predictor, by fine-tuning a pre-trained representation on human collected video sequences. The final method is able to substantially outperform traditional robot learning baselines (e.g. 70% success v.s. 50% for behavior cloning on pick-place) on a suite of diverse real-world manipulation tasks. It can also generalize to novel objects, without using any robot demonstrations during train time. For visualizations of the learned policies please check: https://agi-labs.github.io/manipulate-by-seeing/
In content-based video retrieval (CBVR), dealing with large-scale collections, efficiency is as important as accuracy. For this reason, several video-level feature-based studies have actively been conducted; nevertheless, owing to the severe difficulty of embedding a lengthy and untrimmed video into a single feature, these studies have shown insufficient for accurate retrieval compared to frame-level feature-based studies. In this paper, we show an insight that appropriate suppression of irrelevant frames can be a clue to overcome the current obstacles of the video-level feature-based approaches. Furthermore, we propose a Video-to-Video Suppression network (VVS) as a solution. The VVS is an end-to-end framework that consists of an easy distractor elimination stage for identifying which frames to remove and a suppression weight generation stage for determining how much to suppress the remaining frames. This structure is intended to effectively describe an untrimmed video with varying content and meaningless information. Its efficacy is proved via extensive experiments, and we show that our approach is not only state-of-the-art in video-level feature-based approaches but also has a fast inference time despite possessing retrieval capabilities close to those of frame-level feature-based approaches.
Deep Neural Networks (DNNs) are increasingly applied in the real world in safety critical applications like advanced driver assistance systems. An example for such use case is represented by traffic sign recognition systems. At the same time, it is known that current DNNs can be fooled by adversarial attacks, which raises safety concerns if those attacks can be applied under realistic conditions. In this work we apply different black-box attack methods to generate perturbations that are applied in the physical environment and can be used to fool systems under different environmental conditions. To the best of our knowledge we are the first to combine a general framework for physical attacks with different black-box attack methods and study the impact of the different methods on the success rate of the attack under the same setting. We show that reliable physical adversarial attacks can be performed with different methods and that it is also possible to reduce the perceptibility of the resulting perturbations. The findings highlight the need for viable defenses of a DNN even in the black-box case, but at the same time form the basis for securing a DNN with methods like adversarial training which utilizes adversarial attacks to augment the original training data.
Combining machine learning and constrained optimization, Predict+Optimize tackles optimization problems containing parameters that are unknown at the time of solving. Prior works focus on cases with unknowns only in the objectives. A new framework was recently proposed to cater for unknowns also in constraints by introducing a loss function, called Post-hoc Regret, that takes into account the cost of correcting an unsatisfiable prediction. Since Post-hoc Regret is non-differentiable, the previous work computes only its approximation. While the notion of Post-hoc Regret is general, its specific implementation is applicable to only packing and covering linear programming problems. In this paper, we first show how to compute Post-hoc Regret exactly for any optimization problem solvable by a recursive algorithm satisfying simple conditions. Experimentation demonstrates substantial improvement in the quality of solutions as compared to the earlier approximation approach. Furthermore, we show experimentally the empirical behavior of different combinations of correction and penalty functions used in the Post-hoc Regret of the same benchmarks. Results provide insights for defining the appropriate Post-hoc Regret in different application scenarios.
This paper proposes a deep learning (DL) model for automatic sleep stage classification based on single-channel EEG data. The DL model features a convolutional neural network (CNN) and transformers. The model was designed to run on energy and memory-constrained devices for real-time operation with local processing. The Fpz-Cz EEG signals from a publicly available Sleep-EDF dataset are used to train and test the model. Four convolutional filter layers were used to extract features and reduce the data dimension. Then, transformers were utilized to learn the time-variant features of the data. To improve performance, we also implemented a subject specific training before the inference (i.e., prediction) stage. With the subject specific training, the F1 score was 0.91, 0.37, 0.84, 0.877, and 0.73 for wake, N1-N3, and rapid eye movement (REM) stages, respectively. The performance of the model was comparable to the state-of-the-art works with significantly greater computational costs. We tested a reduced-sized version of the proposed model on a low-cost Arduino Nano 33 BLE board and it was fully functional and accurate. In the future, a fully integrated wireless EEG sensor with edge DL will be developed for sleep research in pre-clinical and clinical experiments, such as real-time sleep modulation.
Real-time monitoring of human behaviours, especially in e-Health applications, has been an active area of research in the past decades. On top of IoT-based sensing environments, anomaly detection algorithms have been proposed for the early detection of abnormalities. Gradual change procedures, commonly referred to as drift anomalies, have received much less attention in the literature because they represent a much more challenging scenario than sudden temporary changes (point anomalies). In this paper, we propose, for the first time, a fully unsupervised real-time drift detection algorithm named DynAmo, which can identify drift periods as they are happening. DynAmo comprises a dynamic clustering component to capture the overall trends of monitored behaviours and a trajectory generation component, which extracts features from the densest cluster centroids. Finally, we apply an ensemble of divergence tests on sliding reference and detection windows to detect drift periods in the behavioural sequence.
Evolutionary computation-based neural architecture search (ENAS) is a popular technique for automating architecture design of deep neural networks. In recent years, various ENAS algorithms have been proposed and shown promising performance on diverse real-world applications. In contrast to these groundbreaking applications, there is no theoretical guideline for assigning a reasonable running time (mainly affected by the generation number, population size, and evolution operator) given both the anticipated performance and acceptable computation budget on ENAS problems. The expected hitting time (EHT), which refers to the average generations, is considered to analyze the running time of ENAS algorithms. This paper proposes a general framework for estimating the EHT of ENAS algorithms, which includes common configuration, search space partition, transition probability estimation, and hitting time analysis. By exploiting the proposed framework, we consider the so-called ($\lambda$+$\lambda$)-ENAS algorithms with different mutation operators and manage to estimate the lower bounds of the EHT {which are critical for the algorithm to find the global optimum}. Furthermore, we study the theoretical results on the NAS-Bench-101 architecture searching problem, and the results show that the one-bit mutation with "bit-based fair mutation" strategy needs less time than the "offspring-based fair mutation" strategy, and the bitwise mutation operator needs less time than the $q$-bit mutation operator. To the best of our knowledge, this is the first work focusing on the theory of ENAS, and the above observation will be substantially helpful in designing efficient ENAS algorithms.
We propose a multisensor fusion framework for onboard real-time navigation of a quadrotor in an indoor environment, by integrating sensor readings from an Inertial Measurement Unit (IMU), a camera-based object detection algorithm, and an Ultra-WideBand (UWB) localization system. The sensor readings from the camera-based object detection algorithm and the UWB localization system arrive intermittently, since the measurements are not readily available. We design a Kalman filter that manages intermittent observations in order to handle and fuse the readings and estimate the pose of the quadrotor for tracking a predefined trajectory. The system is implemented via a Hardware-in-the-loop (HIL) simulation technique, in which the dynamic model of the quadrotor is simulated in an open-source 3D robotics simulator tool, and the whole navigation system is implemented on Artificial Intelligence (AI) enabled edge GPU. The simulation results show that our proposed framework offers low positioning and trajectory errors, while handling intermittent sensor measurements.
Hidden semi-Markov Models (HSMM's) - while broadly in use - are restricted to a discrete and uniform time grid. They are thus not well suited to explain often irregularly spaced discrete event data from continuous-time phenomena. We show that non-sampling-based latent state inference used in HSMM's can be generalized to latent Continuous-Time semi-Markov Chains (CTSMC's). We formulate integro-differential forward and backward equations adjusted to the observation likelihood and introduce an exact integral equation for the Bayesian posterior marginals and a scalable Viterbi-type algorithm for posterior path estimates. The presented equations can be efficiently solved using well-known numerical methods. As a practical tool, variable-step HSMM's are introduced. We evaluate our approaches in latent state inference scenarios in comparison to classical HSMM's.
Generative Adversarial Networks (GAN) have been widely investigated for image synthesis based on their powerful representation learning ability. In this work, we explore the StyleGAN and its application of synthetic food image generation. Despite the impressive performance of GAN for natural image generation, food images suffer from high intra-class diversity and inter-class similarity, resulting in overfitting and visual artifacts for synthetic images. Therefore, we aim to explore the capability and improve the performance of GAN methods for food image generation. Specifically, we first choose StyleGAN3 as the baseline method to generate synthetic food images and analyze the performance. Then, we identify two issues that can cause performance degradation on food images during the training phase: (1) inter-class feature entanglement during multi-food classes training and (2) loss of high-resolution detail during image downsampling. To address both issues, we propose to train one food category at a time to avoid feature entanglement and leverage image patches cropped from high-resolution datasets to retain fine details. We evaluate our method on the Food-101 dataset and show improved quality of generated synthetic food images compared with the baseline. Finally, we demonstrate the great potential of improving the performance of downstream tasks, such as food image classification by including high-quality synthetic training samples in the data augmentation.