Recently, groundbreaking results have been presented on open-vocabulary semantic image segmentation. Such methods segment each pixel in an image into arbitrary categories provided at run-time in the form of text prompts, as opposed to a fixed set of classes defined at training time. In this work, we present a zero-shot volumetric open-vocabulary semantic scene segmentation method. Our method builds on the insight that we can fuse image features from a vision-language model into a neural implicit representation. We show that the resulting feature field can be segmented into different classes by assigning points to natural language text prompts. The implicit volumetric representation enables us to segment the scene both in 3D and 2D by rendering feature maps from any given viewpoint of the scene. We show that our method works on noisy real-world data and can run in real-time on live sensor data dynamically adjusting to text prompts. We also present quantitative comparisons on the ScanNet dataset.

Abstract--We present a demonstration of service-based trajectory planning for a drone delivery system in a multi-drone skyway network. We conduct several experiments using Crazyflie drones to collect the drone's position data, wind speed and direction, and wind effects on voltage consumption rates. The experiments are run for a varying number of recharging stations, wind speed, and wind direction in a multi-drone skyway network. Drones are a specific type of unmanned aerial vehicles that Figure 1: Multi-Drone Skyway Network fly autonomously with full network connectivity capabilities where the drones wait for the recharging pad availability to [1]. This connectivity enables drones to operate safely and be recharged.

Deploying models on target domain data subject to distribution shift requires adaptation. Test-time training (TTT) emerges as a solution to this adaptation under a realistic scenario where access to full source domain data is not available, and instant inference on the target domain is required. Despite many efforts into TTT, there is a confusion over the experimental settings, thus leading to unfair comparisons. In this work, we first revisit TTT assumptions and categorize TTT protocols by two key factors. Among the multiple protocols, we adopt a realistic sequential test-time training (sTTT) protocol, under which we develop a test-time anchored clustering (TTAC) approach to enable stronger test-time feature learning. TTAC discovers clusters in both source and target domains and matches the target clusters to the source ones to improve adaptation. When source domain information is strictly absent (i.e. source-free) we further develop an efficient method to infer source domain distributions for anchored clustering. Finally, self-training~(ST) has demonstrated great success in learning from unlabeled data and we empirically figure out that applying ST alone to TTT is prone to confirmation bias. Therefore, a more effective TTT approach is introduced by regularizing self-training with anchored clustering, and the improved model is referred to as TTAC++. We demonstrate that, under all TTT protocols, TTAC++ consistently outperforms the state-of-the-art methods on five TTT datasets, including corrupted target domain, selected hard samples, synthetic-to-real adaptation and adversarially attacked target domain. We hope this work will provide a fair benchmarking of TTT methods, and future research should be compared within respective protocols.

Aerial scene classification, which aims to semantically label remote sensing images in a set of predefined classes (e.g., agricultural, beach, and harbor), is a very challenging task in remote sensing due to high intra-class variability and the different scales and orientations of the objects present in the dataset images. In remote sensing area, the use of CNN architectures as an alternative solution is also a reality for scene classification tasks. Generally, these CNNs are used to perform the traditional image classification task. However, another less used way to classify remote sensing image might be the one that uses deep metric learning (DML) approaches. In this sense, this work proposes to employ six DML approaches for aerial scene classification tasks, analysing their behave with four different pre-trained CNNs as well as combining them through the use of evolutionary computation algorithm (UMDA). In performed experiments, it is possible to observe than DML approaches can achieve the best classification results when compared to traditional pre-trained CNNs for three well-known remote sensing aerial scene datasets. In addition, the UMDA algorithm proved to be a promising strategy to combine DML approaches when there is diversity among them, managing to improve at least 5.6% of accuracy in the classification results using almost 50\% of the available classifiers for the construction of the final ensemble of classifiers.

End-to-end speech recognition models are improved by incorporating external text sources, typically by fusion with an external language model. Such language models have to be retrained whenever the corpus of interest changes. Furthermore, since they store the entire corpus in their parameters, rare words can be challenging to recall. In this work, we propose augmenting a transducer-based ASR model with a retrieval language model, which directly retrieves from an external text corpus plausible completions for a partial ASR hypothesis. These completions are then integrated into subsequent predictions by an adapter, which is trained once, so that the corpus of interest can be switched without incurring the computational overhead of retraining. Our experiments show that the proposed model significantly improves the performance of a transducer baseline on a pair of question-answering datasets. Further, it outperforms shallow fusion on recognition of named entities by about 7 relative; when the two are combined, the relative improvement increases to 13%.

In complex large-scale systems such as climate, important effects are caused by a combination of confounding processes that are not fully observable. The identification of sources from observations of system state is vital for attribution and prediction, which inform critical policy decisions. The difficulty of these types of inverse problems lies in the inability to isolate sources and the cost of simulating computational models. Surrogate models may enable the many-query algorithms required for source identification, but data challenges arise from high dimensionality of the state and source, limited ensembles of costly model simulations to train a surrogate model, and few and potentially noisy state observations for inversion due to measurement limitations. The influence of auxiliary processes adds an additional layer of uncertainty that further confounds source identification. We introduce a framework based on (1) calibrating deep neural network surrogates to the flow maps provided by an ensemble of simulations obtained by varying sources, and (2) using these surrogates in a Bayesian framework to identify sources from observations via optimization. Focusing on an atmospheric dispersion exemplar, we find that the expressive and computationally efficient nature of the deep neural network operator surrogates in appropriately reduced dimension allows for source identification with uncertainty quantification using limited data. Introducing a variable wind field as an auxiliary process, we find that a Bayesian approximation error approach is essential for reliable source inversion when uncertainty due to wind stresses the algorithm.

Meta-learning, decision fusion, hybrid models, and representation learning are topics of investigation with significant traction in time-series forecasting research. Of these two specific areas have shown state-of-the-art results in forecasting: hybrid meta-learning models such as Exponential Smoothing - Recurrent Neural Network (ES-RNN) and Neural Basis Expansion Analysis (N-BEATS) and feature-based stacking ensembles such as Feature-based FORecast Model Averaging (FFORMA). However, a unified taxonomy for model fusion and an empirical comparison of these hybrid and feature-based stacking ensemble approaches is still missing. This study presents a unified taxonomy encompassing these topic areas. Furthermore, the study empirically evaluates several model fusion approaches and a novel combination of hybrid and feature stacking algorithms called Deep-learning FORecast Model Averaging (DeFORMA). The taxonomy contextualises the considered methods. Furthermore, the empirical analysis of the results shows that the proposed model, DeFORMA, can achieve state-of-the-art results in the M4 data set. DeFORMA, increases the mean Overall Weighted Average (OWA) in the daily, weekly and yearly subsets with competitive results in the hourly, monthly and quarterly subsets. The taxonomy and empirical results lead us to argue that significant progress is still to be made by continuing to explore the intersection of these research areas.

Transformer is a transformative framework for deep learning which models sequential data and has achieved remarkable performance on a wide range of tasks, but with high computational and energy cost. To improve its efficiency, a popular choice is to compress the models via binarization which constrains the floatingpoint values into binary ones to save resource consumption owing to cheap bitwise operations significantly. However, existing binarization methods only aim at minimizing the information loss for the input distribution statistically, while ignoring the pairwise similarity modeling at the core of the attention mechanism. To this end, we propose a new binarization paradigm customized to high-dimensional softmax attention via kernelized hashing, called EcoFormer, to map the original queries and keys into low-dimensional binary codes in Hamming space. The kernelized hash functions are learned to match the ground-truth similarity relations extracted from the attention map in a self-supervised way. Based on the equivalence between the inner product of binary codes and the Hamming distance as well as the associative property of matrix multiplication, we can approximate the attention in linear complexity by expressing it as a dot-product of binary codes. Moreover, the compact binary representations of queries and keys in EcoFormer enable us to replace most of the expensive multiply-accumulate operations in attention with simple accumulations to save considerable on-chip energy footprint on edge devices. Extensive experiments on both vision and language tasks show that EcoFormer consistently achieves comparable performance with standard attentions while consuming much fewer resources. For example, based on PVTv2-B0 and ImageNet-1K, EcoFormer achieves a 73% reduction in on-chip energy footprint with only a slight performance drop of 0.33% compared to the standard attention.

We establish a layer-wise parameterization for 1D convolutional neural networks (CNNs) with built-in end-to-end robustness guarantees. Herein, we use the Lipschitz constant of the input-output mapping characterized by a CNN as a robustness measure. We base our parameterization on the Cayley transform that parameterizes orthogonal matrices and the controllability Gramian for the state space representation of the convolutional layers. The proposed parameterization by design fulfills linear matrix inequalities that are sufficient for Lipschitz continuity of the CNN, which further enables unconstrained training of Lipschitz-bounded 1D CNNs. Finally, we train Lipschitz-bounded 1D CNNs for the classification of heart arrythmia data and show their improved robustness.

Counterfactual fairness is an approach to AI fairness that tries to make decisions based on the outcomes that an individual with some kind of sensitive status would have had without this status. This paper proposes Double Machine Learning (DML) Fairness which analogises this problem of counterfactual fairness in regression problems to that of estimating counterfactual outcomes in causal inference under the Potential Outcomes framework. It uses arbitrary machine learning methods to partial out the effect of sensitive variables on nonsensitive variables and outcomes. Assuming that the effects of the two sets of variables are additively separable, outcomes will be approximately equalised and individual-level outcomes will be counterfactually fair. This paper demonstrates the approach in a simulation study pertaining to discrimination in workplace hiring and an application on real data estimating the GPAs of law school students. It then discusses when it is appropriate to apply such a method to problems of real-world discrimination where constructs are conceptually complex and finally, whether DML Fairness can achieve justice in these settings.