Search

We introduce a variant of Error Correcting Codes with no predetermined length. An Unbounded ECC with rate $R$ and distance $\varepsilon$ is an encoding of a possibly infinite message into a possibly infinite codeword, such that for every large enough $k$ we may recover the first $Rk$ symbols of the message from the first $k$ symbols of the codeword -- even when up to $\frac{1}{2}\varepsilon k$ of these codeword symbols are adversarially corrupted. We study unbounded codes over a binary alphabet in the regime of small distance $\varepsilon$, and obtain nearly-tight upper and lower bounds in several natural settings. We show that the optimal rate of such a code is between $R<1-\Omega(\sqrt{\varepsilon})$ and $R>1-O\left(\sqrt{\varepsilon\log\log\left(1/\varepsilon\right)}\right)$. Surprisingly, our construction is non-linear, and we show that the optimal rate of a linear unbounded code is the asymptotically worse $R=1-\Theta\left(\sqrt{\varepsilon\log\left(1/\varepsilon\right)}\right)$. In the setting of random noise, the optimal rate of unbounded codes improves and matches the rate of standard codes at $R=1-\Theta({\varepsilon\log{\left(1/\varepsilon\right)}})$.

The inherent trade-off in on-line learning is between exploration and exploitation. A good balance between these two (conflicting) goals can achieve a better long-term performance. Can we define an optimal balance? We propose to study this question through a backward-adaptive lossy compression system, which exhibits a "natural" trade-off between exploration and exploitation., Comment: This paper was presented as a poster in the workshop `Learn 2 Compress', in ISIT 2024, Athens, Greece, July 2024. It was processed and reviewed in the Open Review system

Estimating the second frequency moment of a stream up to $(1\pm\varepsilon)$ multiplicative error requires at most $O(\log n / \varepsilon^2)$ bits of space, due to a seminal result of Alon, Matias, and Szegedy. It is also known that at least $\Omega(\log n + 1/\varepsilon^{2})$ space is needed. We prove an optimal lower bound of $\Omega\left(\log \left(n \varepsilon^2 \right) / \varepsilon^2\right)$ for all $\varepsilon = \Omega(1/\sqrt{n})$. Note that when $\varepsilon>n^{-1/2 + c}$, where $c>0$, our lower bound matches the classic upper bound of AMS. For smaller values of $\varepsilon$ we also introduce a revised algorithm that improves the classic AMS bound and matches our lower bound. Our lower bound holds also for the more general problem of $p$-th frequency moment estimation for the range of $p\in (1,2]$, giving a tight bound in the only remaining range to settle the optimal space complexity of estimating frequency moments.

Blindly estimating the direction of arrival (DoA) of early room reflections without prior knowledge of the room impulse response or source signal is highly valuable in audio signal processing applications. The FF-PHALCOR (Frequency Focusing PHase ALigned CORrelation) method was recently developed for this purpose, extending the original PHALCOR method to work with arbitrary arrays rather than just spherical ones. Previous studies have provided only initial insights into its performance. This study offers a comprehensive analysis of the method's performance and limitations, examining how reflection characteristics such as delay, amplitude, and spatial density affect its effectiveness. The research also proposes improvements to overcome these limitations, enhancing detection quality and reducing false alarms. Additionally, the study examined how spatial perception is affected by generating room impulse responses using estimated reflection information. The findings suggest a perceptual advantage of the proposed approach over the baseline, with particularly high perceptual quality when using the spherical array with 32 microphones. However, the quality is somewhat reduced when using a semi-circular array with only 6 microphones.

The increasing popularity of spatial audio in applications such as teleconferencing, entertainment, and virtual reality has led to the recent developments of binaural reproduction methods. However, only a few of these methods are well-suited for wearable and mobile arrays, which typically consist of a small number of microphones. One such method is binaural signal matching (BSM), which has been shown to produce high-quality binaural signals for wearable arrays. However, BSM may be suboptimal in cases of high direct-to-reverberant ratio (DRR) as it is based on the diffuse sound field assumption. To overcome this limitation, previous studies incorporated sound-field models other than diffuse. However, this approach was not studied comprehensively. This paper extensively investigates two BSM-based methods designed for high DRR scenarios. The methods incorporate a sound field model composed of direct and reverberant components.The methods are investigated both mathematically and using simulations, finally validated by a listening test. The results show that the proposed methods can significantly improve the performance of BSM , in particular in the direction of the source, while presenting only a negligible degradation in other directions. Furthermore, when source direction estimation is inaccurate, performance of these methods degrade to equal that of the BSM, presenting a desired robustness quality.

Scheduling in multi-channel wireless communication system presents formidable challenges in effectively allocating resources. To address these challenges, we investigate a multi-resource restless matching bandit (MR-RMB) model for heterogeneous resource systems with an objective of maximizing long-term discounted total rewards while respecting resource constraints. We have also generalized to applications beyond multi-channel wireless. We discuss the Max-Weight Index Matching algorithm, which optimizes resource allocation based on learned partial indexes. We have derived the policy gradient theorem for index learning. Our main contribution is the introduction of a new Deep Index Policy (DIP), an online learning algorithm tailored for MR-RMB. DIP learns the partial index by leveraging the policy gradient theorem for restless arms with convoluted and unknown transition kernels of heterogeneous resources. We demonstrate the utility of DIP by evaluating its performance for three different MR-RMB problems. Our simulation results show that DIP indeed learns the partial indexes efficiently.

This study investigates the approach of direction-dependent selection of Head-Related Transfer Functions (HRTFs) and its impact on sound localization accuracy. For applications such as virtual reality (VR) and teleconferencing, obtaining individualized HRTFs can be beneficial yet challenging, the objective of this work is therefore to assess whether incorporating HRTFs in a direction-dependent manner could improve localization precision without the need to obtain individualized HRTFs. A localization experiment conducted with a VR headset assessed localization errors, comparing an overall best HRTF from a set, against selecting the best HRTF based on average performance in each direction. The results demonstrate a substantial improvement in elevation localization error with the method motivated by direction-dependent HRTF selection, while revealing insignificant differences in azimuth errors., Comment: Accepted for publication in the 2024 AES International Conference on Audio for Virtual and Augmented Reality, 5 pages, 4 figures

Binaural reproduction for headphone-centric listening has become a focal point in ongoing research, particularly within the realm of advancing technologies such as augmented and virtual reality (AR and VR). The demand for high-quality spatial audio in these applications is essential to uphold a seamless sense of immersion. However, challenges arise from wearable recording devices equipped with only a limited number of microphones and irregular microphone placements due to design constraints. These factors contribute to limited reproduction quality compared to reference signals captured by high-order microphone arrays. This paper introduces a novel optimization loss tailored for a beamforming-based, signal-independent binaural reproduction scheme. This method, named iMagLS-BSM incorporates an interaural level difference (ILD) error term into the previously proposed binaural signal matching (BSM) magnitude least squares (MagLS) rendering loss for lateral plane angles. The method leverages nonlinear programming to minimize the introduced loss. Preliminary results show a substantial reduction in ILD error, while maintaining a binaural magnitude error comparable to that achieved with a MagLS BSM solution. These findings hold promise for enhancing the overall spatial quality of resultant binaural signals., Comment: Paper accepted for publication in IWAENC 2024, 4 pages, 2 figures

Binaural reproduction is rapidly becoming a topic of great interest in the research community, especially with the surge of new and popular devices, such as virtual reality headsets, smart glasses, and head-tracked headphones. In order to immerse the listener in a virtual or remote environment with such devices, it is essential to generate realistic and accurate binaural signals. This is challenging, especially since the microphone arrays mounted on these devices are typically composed of an arbitrarily-arranged small number of microphones, which impedes the use of standard audio formats like Ambisonics, and provides limited spatial resolution. The binaural signal matching (BSM) method was developed recently to overcome these challenges. While it produced binaural signals with low error using relatively simple arrays, its performance degraded significantly when head rotation was introduced. This paper aims to develop the BSM method further and overcome its limitations. For this purpose, the method is first analyzed in detail, and a design framework that guarantees accurate binaural reproduction for relatively complex acoustic environments is presented. Next, it is shown that the BSM accuracy may significantly degrade at high frequencies, and thus, a perceptually motivated extension to the method is proposed, based on a magnitude least-squares (MagLS) formulation. These insights and developments are then analyzed with the help of an extensive simulation study of a simple six-microphone semi-circular array. It is further shown that the BSM-MagLS method can be very useful in compensating for head rotations with this array. Finally, a listening experiment is conducted with a four-microphone array on a pair of glasses in a reverberant speech environment and including head rotations, where it is shown that BSM-MagLS can indeed produce binaural signals with a high perceived quality., Comment: Submitted to EURASIP Journal on audio speech and music processing

Different users find different images generated for the same prompt desirable. This gives rise to personalized image generation which involves creating images aligned with an individual's visual preference. Current generative models are, however, unpersonalized, as they are tuned to produce outputs that appeal to a broad audience. Using them to generate images aligned with individual users relies on iterative manual prompt engineering by the user which is inefficient and undesirable. We propose to personalize the image generation process by first capturing the generic preferences of the user in a one-time process by inviting them to comment on a small selection of images, explaining why they like or dislike each. Based on these comments, we infer a user's structured liked and disliked visual attributes, i.e., their visual preference, using a large language model. These attributes are used to guide a text-to-image model toward producing images that are tuned towards the individual user's visual preference. Through a series of user studies and large language model guided evaluations, we demonstrate that the proposed method results in generations that are well aligned with individual users' visual preferences., Comment: Project page at https://viper.epfl.ch/

A de facto standard in solving computer vision problems is to use a common high-resolution camera and choose its placement on an agent (i.e., position and orientation) based on human intuition. On the other hand, extremely simple and well-designed visual sensors found throughout nature allow many organisms to perform diverse, complex behaviors. In this work, motivated by these examples, we raise the following questions: 1. How effective simple visual sensors are in solving vision tasks? 2. What role does their design play in their effectiveness? We explore simple sensors with resolutions as low as one-by-one pixel, representing a single photoreceptor First, we demonstrate that just a few photoreceptors can be enough to solve many tasks, such as visual navigation and continuous control, reasonably well, with performance comparable to that of a high-resolution camera. Second, we show that the design of these simple visual sensors plays a crucial role in their ability to provide useful information and successfully solve these tasks. To find a well-performing design, we present a computational design optimization algorithm and evaluate its effectiveness across different tasks and domains, showing promising results. Finally, we perform a human survey to evaluate the effectiveness of intuitive designs devised manually by humans, showing that the computationally found design is among the best designs in most cases.

Current multimodal and multitask foundation models like 4M or UnifiedIO show promising results, but in practice their out-of-the-box abilities to accept diverse inputs and perform diverse tasks are limited by the (usually rather small) number of modalities and tasks they are trained on. In this paper, we expand upon the capabilities of them by training a single model on tens of highly diverse modalities and by performing co-training on large-scale multimodal datasets and text corpora. This includes training on several semantic and geometric modalities, feature maps from recent state of the art models like DINOv2 and ImageBind, pseudo labels of specialist models like SAM and 4DHumans, and a range of new modalities that allow for novel ways to interact with the model and steer the generation, for example image metadata or color palettes. A crucial step in this process is performing discrete tokenization on various modalities, whether they are image-like, neural network feature maps, vectors, structured data like instance segmentation or human poses, or data that can be represented as text. Through this, we expand on the out-of-the-box capabilities of multimodal models and specifically show the possibility of training one model to solve at least 3x more tasks/modalities than existing ones and doing so without a loss in performance. This enables more fine-grained and controllable multimodal generation capabilities and allows us to study the distillation of models trained on diverse data and objectives into a unified model. We successfully scale the training to a three billion parameter model using tens of modalities and different datasets. The resulting models and training code are open sourced at 4m.epfl.ch., Comment: Project page at 4m.epfl.ch

Over an algebraically closed field, the $\textit{double point interpolation}$ problem asks for the vector space dimension of the projective hypersurfaces of degree $d$ singular at a given set of points. After being open for 90 years, a series of papers by J. Alexander and A. Hirschowitz in 1992--1995 settled this question in what is referred to as the Alexander-Hirschowitz theorem. In this paper we primarily use commutative algebra to lay the groundwork necessary to prove analogous statements in the $\textit{weighted projective space}$, a natural generalization of the projective space. We show the Hilbert function of general simple points in any $n$-dimensional weighted projective space exhibits the expected behavior. We give an inductive procedure for weighted projective space, similar to that originally due to A. Terracini from 1915, to demonstrate an example of a weighted projective plane where the analogue of the Alexander-Hirschowitz theorem holds without exceptions. We further adapt Terracini's lemma regarding secant varieties to give an interpolation bound for an infinite family of weighted projective planes., Comment: 40 pages, 2 figures. Comments are welcomed. It was shown that P(1,2,3) is the only weighted projective plane with no exceptions. The definition of AH_n(d) was updated. Some number of typos were fixed

Analog codes add redundancy by expanding the dimension using real/complex-valued operations. Frame theory provides a mathematical basis for constructing such codes, with diverse applications in non-orthogonal code-division multiple access (NOMA-CDMA), distributed computation, multiple description source coding, space-time coding (STC), and more. The channel model corresponding to these applications is a combination of noise and erasures. Recent analyses showed a useful connection between spectral random-matrix theory and large equiangular tight frames (ETFs) under random uniform erasures. In this work we generalize this model to a channel where the erasures come in blocks. This particularly fits NOMA-CDMA with multiple transmit antennas for each user and STC with known spatial grouping. We present a method to adjust ETF codes to suit block erasures, and find minimum intra-block-correlation frames which outperform ETFs in this setting.

Over the last decade, similar to other application domains, social media content has been proven very effective in disaster informatics. However, due to the unstructured nature of the data, several challenges are associated with disaster analysis in social media content. To fully explore the potential of social media content in disaster informatics, access to relevant content and the correct geo-location information is very critical. In this paper, we propose a three-step solution to tackling these challenges. Firstly, the proposed solution aims to classify social media posts into relevant and irrelevant posts followed by the automatic extraction of location information from the posts' text through Named Entity Recognition (NER) analysis. Finally, to quickly analyze the topics covered in large volumes of social media posts, we perform topic modeling resulting in a list of top keywords, that highlight the issues discussed in the tweet. For the Relevant Classification of Twitter Posts (RCTP), we proposed a merit-based fusion framework combining the capabilities of four different models namely BERT, RoBERTa, Distil BERT, and ALBERT obtaining the highest F1-score of 0.933 on a benchmark dataset. For the Location Extraction from Twitter Text (LETT), we evaluated four models namely BERT, RoBERTa, Distil BERTA, and Electra in an NER framework obtaining the highest F1-score of 0.960. For topic modeling, we used the BERTopic library to discover the hidden topic patterns in the relevant tweets. The experimental results of all the components of the proposed end-to-end solution are very encouraging and hint at the potential of social media content and NLP in disaster management., Comment: 15 pages; 4 tables; 4 figures

This paper focuses on a very important societal challenge of water quality analysis. Being one of the key factors in the economic and social development of society, the provision of water and ensuring its quality has always remained one of the top priorities of public authorities. To ensure the quality of water, different methods for monitoring and assessing the water networks, such as offline and online surveys, are used. However, these surveys have several limitations, such as the limited number of participants and low frequency due to the labor involved in conducting such surveys. In this paper, we propose a Natural Language Processing (NLP) framework to automatically collect and analyze water-related posts from social media for data-driven decisions. The proposed framework is composed of two components, namely (i) text classification, and (ii) topic modeling. For text classification, we propose a merit-fusion-based framework incorporating several Large Language Models (LLMs) where different weight selection and optimization methods are employed to assign weights to the LLMs. In topic modeling, we employed the BERTopic library to discover the hidden topic patterns in the water-related tweets. We also analyzed relevant tweets originating from different regions and countries to explore global, regional, and country-specific issues and water-related concerns. We also collected and manually annotated a large-scale dataset, which is expected to facilitate future research on the topic., Comment: 11 pages, 6 figures, and 3 tables

Extreme rainfall in the Indian summer monsoon can be destructive and deadly. Although El Ni\~no/ events in the equatorial Pacific make dry days and whole summers more likely throughout India, their influence on daily extremes is not well established. Despite this summer-mean drying effect, we show using observational data spanning 1901-2020 that El Ni\~no increases extreme rainfall likelihoods within monsoonal India, especially in the the summer's core rainy areas of central-eastern India and the narrow southwestern coastal band. Conversely, extremes are broadly suppressed in the drier southeast and far northwest, and more moderate accumulations are inhibited throughout the domain. These rainfall signals appear driven by corresponding ones in convective buoyancy, provided both the undilute instability of near-surface air and its dilution by mixing with drier air above are accounted for. When the summer ENSO state is predicted from a seasonal forecast ensemble initialized in May, the extreme rainfall patterns broadly persist, suggesting the potential for skillful seasonal forecasts. The framework of analyzing the full distributions of rainfall and convective buoyancy could be usefully applied to hourly extremes, other tropical regions under ENSO, other variability modes, and to trends in extreme rainfall under climate change., Comment: 43 pages, 13 figures

Vision-language models (VLMs) are typically composed of a vision encoder, e.g. CLIP, and a language model (LM) that interprets the encoded features to solve downstream tasks. Despite remarkable progress, VLMs are subject to several shortcomings due to the limited capabilities of vision encoders, e.g. "blindness" to certain image features, visual hallucination, etc. To address these issues, we study broadening the visual encoding capabilities of VLMs. We first comprehensively benchmark several vision encoders with different inductive biases for solving VLM tasks. We observe that there is no single encoding configuration that consistently achieves top performance across different tasks, and encoders with different biases can perform surprisingly similarly. Motivated by this, we introduce a method, named BRAVE, that consolidates features from multiple frozen encoders into a more versatile representation that can be directly fed as the input to a frozen LM. BRAVE achieves state-of-the-art performance on a broad range of captioning and VQA benchmarks and significantly reduces the aforementioned issues of VLMs, while requiring a smaller number of trainable parameters than existing methods and having a more compressed representation. Our results highlight the potential of incorporating different visual biases for a more broad and contextualized visual understanding of VLMs., Comment: Project page at https://brave-vlms.epfl.ch/

All-in-one image restoration tackles different types of degradations with a unified model instead of having task-specific, non-generic models for each degradation. The requirement to tackle multiple degradations using the same model can lead to high-complexity designs with fixed configuration that lack the adaptability to more efficient alternatives. We propose DyNet, a dynamic family of networks designed in an encoder-decoder style for all-in-one image restoration tasks. Our DyNet can seamlessly switch between its bulkier and lightweight variants, thereby offering flexibility for efficient model deployment with a single round of training. This seamless switching is enabled by our weights-sharing mechanism, forming the core of our architecture and facilitating the reuse of initialized module weights. Further, to establish robust weights initialization, we introduce a dynamic pre-training strategy that trains variants of the proposed DyNet concurrently, thereby achieving a 50% reduction in GPU hours. Our dynamic pre-training strategy eliminates the need for maintaining separate checkpoints for each variant, as all models share a common set of checkpoints, varying only in model depth. This efficient strategy significantly reduces storage overhead and enhances adaptability. To tackle the unavailability of large-scale dataset required in pre-training, we curate a high-quality, high-resolution image dataset named Million-IRD, having 2M image samples. We validate our DyNet for image denoising, deraining, and dehazing in all-in-one setting, achieving state-of-the-art results with 31.34\% reduction in GFlops and a 56.75\% reduction in parameters compared to baseline models. The source codes and trained models are available at https://github.com/akshaydudhane16/DyNet., Comment: This version includes updates where the DyNet variants now share the same weights during inference as well, eliminating the need to store separate weights and thereby reducing device storage requirements. Additionally, all results have been updated based on the new experimental setup

A subset $S$ of the Boolean hypercube $\mathbb{F}_2^n$ is a sumset if $S = A+A = \{a + b \ | \ a, b\in A\}$ for some $A \subseteq \mathbb{F}_2^n$. We prove that the number of sumsets in $\mathbb{F}_2^n$ is asymptotically $(2^n-1)2^{2^{n-1}}$. Furthermore, we show that the family of sumsets in $\mathbb{F}_2^n$ is almost identical to the family of all subsets of $\mathbb{F}_2^n$ that contain a complete linear subspace of co-dimension $1$.

In this work, we present a method to control a text-to-image generative model to produce training data specifically "useful" for supervised learning. Unlike previous works that employ an open-loop approach and pre-define prompts to generate new data using either a language model or human expertise, we develop an automated closed-loop system which involves two feedback mechanisms. The first mechanism uses feedback from a given supervised model and finds adversarial prompts that result in image generations that maximize the model loss. While these adversarial prompts result in diverse data informed by the model, they are not informed of the target distribution, which can be inefficient. Therefore, we introduce the second feedback mechanism that guides the generation process towards a certain target distribution. We call the method combining these two mechanisms Guided Adversarial Prompts. We perform our evaluations on different tasks, datasets and architectures, with different types of distribution shifts (spuriously correlated data, unseen domains) and demonstrate the efficiency of the proposed feedback mechanisms compared to open-loop approaches., Comment: Project page at https://adversarial-prompts.epfl.ch/

In the image acquisition process, various forms of degradation, including noise, haze, and rain, are frequently introduced. These degradations typically arise from the inherent limitations of cameras or unfavorable ambient conditions. To recover clean images from degraded versions, numerous specialized restoration methods have been developed, each targeting a specific type of degradation. Recently, all-in-one algorithms have garnered significant attention by addressing different types of degradations within a single model without requiring prior information of the input degradation type. However, these methods purely operate in the spatial domain and do not delve into the distinct frequency variations inherent to different degradation types. To address this gap, we propose an adaptive all-in-one image restoration network based on frequency mining and modulation. Our approach is motivated by the observation that different degradation types impact the image content on different frequency subbands, thereby requiring different treatments for each restoration task. Specifically, we first mine low- and high-frequency information from the input features, guided by the adaptively decoupled spectra of the degraded image. The extracted features are then modulated by a bidirectional operator to facilitate interactions between different frequency components. Finally, the modulated features are merged into the original input for a progressively guided restoration. With this approach, the model achieves adaptive reconstruction by accentuating the informative frequency subbands according to different input degradations. Extensive experiments demonstrate that the proposed method achieves state-of-the-art performance on different image restoration tasks, including denoising, dehazing, deraining, motion deblurring, and low-light image enhancement. Our code is available at https://github.com/c-yn/AdaIR., Comment: 28 pages,15 figures

High fidelity spatial audio often performs better when produced using a personalized head-related transfer function (HRTF). However, the direct acquisition of HRTFs is cumbersome and requires specialized equipment. Thus, many personalization methods estimate HRTF features from easily obtained anthropometric features of the pinna, head, and torso. The first HRTF notch frequency (N1) is known to be a dominant feature in elevation localization, and thus a useful feature for HRTF personalization. This paper describes the prediction of N1 frequency from pinna anthropometry using a neural model. Prediction is performed separately on three databases, both simulated and measured, and then by domain mixing in-between the databases. The model successfully predicts N1 frequency for individual databases and by domain mixing between some databases. Prediction errors are better or comparable to those previously reported, showing significant improvement when acquired over a large database and with a larger output range.

In the rapidly evolving fields of virtual and augmented reality, accurate spatial audio capture and reproduction are essential. For these applications, Ambisonics has emerged as a standard format. However, existing methods for encoding Ambisonics signals from arbitrary microphone arrays face challenges, such as errors due to the irregular array configurations and limited spatial resolution resulting from a typically small number of microphones. To address these limitations and challenges, a mathematical framework for studying Ambisonics encoding is presented, highlighting the importance of incorporating the full steering function, and providing a novel measure for predicting the accuracy of encoding each Ambisonics channel from the steering functions alone. Furthermore, novel residual channels are formulated supplementing the Ambisonics channels. A simulation study for several array configurations demonstrates a reduction in binaural error for this approach., Comment: Accepted for presentation at HSCMA 2024

We introduce a cryptographic method to hide an arbitrary secret payload in the response of a Large Language Model (LLM). A secret key is required to extract the payload from the model's response, and without the key it is provably impossible to distinguish between the responses of the original LLM and the LLM that hides a payload. In particular, the quality of generated text is not affected by the payload. Our approach extends a recent result of Christ, Gunn and Zamir (2023) who introduced an undetectable watermarking scheme for LLMs.

A subset $S$ of the Boolean hypercube $\mathbb{F}_2^n$ is a sumset if $S = \{a + b : a, b\in A\}$ for some $A \subseteq \mathbb{F}_2^n$. Sumsets are central objects of study in additive combinatorics, featuring in several influential results. We prove a lower bound of $\Omega(2^{n/2})$ for the number of queries needed to test whether a Boolean function $f:\mathbb{F}_2^n \to \{0,1\}$ is the indicator function of a sumset. Our lower bound for testing sumsets follows from sharp bounds on the related problem of shift testing, which may be of independent interest. We also give a near-optimal $2^{n/2} \cdot \mathrm{poly}(n)$-query algorithm for a smoothed analysis formulation of the sumset refutation problem., Comment: 18 pages

In recent years, the increasing use of Artificial Intelligence based text generation tools has posed new challenges in document provenance, authentication, and authorship detection. However, advancements in stylometry have provided opportunities for automatic authorship and author change detection in multi-authored documents using style analysis techniques. Style analysis can serve as a primary step toward document provenance and authentication through authorship detection. This paper investigates three key tasks of style analysis: (i) classification of single and multi-authored documents, (ii) single change detection, which involves identifying the point where the author switches, and (iii) multiple author-switching detection in multi-authored documents. We formulate all three tasks as classification problems and propose a merit-based fusion framework that integrates several state-of-the-art natural language processing (NLP) algorithms and weight optimization techniques. We also explore the potential of special characters, which are typically removed during pre-processing in NLP applications, on the performance of the proposed methods for these tasks by conducting extensive experiments on both cleaned and raw datasets. Experimental results demonstrate significant improvements over existing solutions for all three tasks on a benchmark dataset., Comment: Tables 7, pages 4

Supervised learning datasets may contain multiple cues that explain the training set equally well, i.e., learning any of them would lead to the correct predictions on the training data. However, many of them can be spurious, i.e., lose their predictive power under a distribution shift and consequently fail to generalize to out-of-distribution (OOD) data. Recently developed "diversification" methods (Lee et al., 2023; Pagliardini et al., 2023) approach this problem by finding multiple diverse hypotheses that rely on different features. This paper aims to study this class of methods and identify the key components contributing to their OOD generalization abilities. We show that (1) diversification methods are highly sensitive to the distribution of the unlabeled data used for diversification and can underperform significantly when away from a method-specific sweet spot. (2) Diversification alone is insufficient for OOD generalization. The choice of the used learning algorithm, e.g., the model's architecture and pretraining, is crucial. In standard experiments (classification on Waterbirds and Office-Home datasets), using the second-best choice leads to an up to 20\% absolute drop in accuracy. (3) The optimal choice of learning algorithm depends on the unlabeled data and vice versa i.e. they are co-dependent. (4) Finally, we show that, in practice, the above pitfalls cannot be alleviated by increasing the number of diverse hypotheses, the major feature of diversification methods. These findings provide a clearer understanding of the critical design factors influencing the OOD generalization abilities of diversification methods. They can guide practitioners in how to use the existing methods best and guide researchers in developing new, better ones., Comment: ICLR 2024

Current machine learning models for vision are often highly specialized and limited to a single modality and task. In contrast, recent large language models exhibit a wide range of capabilities, hinting at a possibility for similarly versatile models in computer vision. In this paper, we take a step in this direction and propose a multimodal training scheme called 4M. It consists of training a single unified Transformer encoder-decoder using a masked modeling objective across a wide range of input/output modalities - including text, images, geometric, and semantic modalities, as well as neural network feature maps. 4M achieves scalability by unifying the representation space of all modalities through mapping them into discrete tokens and performing multimodal masked modeling on a small randomized subset of tokens. 4M leads to models that exhibit several key capabilities: (1) they can perform a diverse set of vision tasks out of the box, (2) they excel when fine-tuned for unseen downstream tasks or new input modalities, and (3) they can function as a generative model that can be conditioned on arbitrary modalities, enabling a wide variety of expressive multimodal editing capabilities with remarkable flexibility. Through experimental analyses, we demonstrate the potential of 4M for training versatile and scalable foundation models for vision tasks, setting the stage for further exploration in multimodal learning for vision and other domains., Comment: NeurIPS 2023 Spotlight. Project page at https://4m.epfl.ch/

Binaural reproduction for headphone-based listening is an active research area due to its widespread use in evolving technologies such as augmented and virtual reality (AR and VR). On the one hand, these applications demand high quality spatial audio perception to preserve the sense of immersion. On the other hand, recording devices may only have a few microphones, leading to low-order representations such as first-order Ambisonics (FOA). However, first-order Ambisonics leads to limited externalization and spatial resolution. In this paper, a novel head-related transfer function (HRTF) preprocessing optimization loss is proposed, and is minimized using nonlinear programming. The new method, denoted iMagLS, involves the introduction of an interaural level difference (ILD) error term to the now widely used MagLS optimization loss for the lateral plane angles. Results indicate that the ILD error could be substantially reduced, while the HRTF magnitude error remains similar to that obtained with MagLS. These results could prove beneficial to the overall spatial quality of first-order Ambisonics, while other reproduction methods could also benefit from considering this modified loss., Comment: 3 pages, 2 figures, Forum Acusticum 2023

The capture and reproduction of spatial audio is becoming increasingly popular, with the mushrooming of applications in teleconferencing, entertainment and virtual reality. Many binaural reproduction methods have been developed and studied extensively for spherical and other specially designed arrays. However, the recent increased popularity of wearable and mobile arrays requires the development of binaural reproduction methods for these arrays. One such method is binaural signal matching (BSM). However, to date this method has only been investigated with fixed matched filters designed for long audio recordings. With the aim of making the BSM method more adaptive to dynamic environments, this paper analyzes BSM with a parameterized sound-field in the time-frequency domain. The paper presents results of implementing the BSM method on a sound-field that was decomposed into its direct and reverberant components, and compares this implementation with the BSM computed for the entire sound-field, to compare performance for binaural reproduction of reverberant speech in a simulated environment.