arXiv subjects

Roger Wattenhofer

Publications and source records attributed to Roger Wattenhofer.

At least 145 records · Page 8Linked to original sources

arXiv2023

Stable Dinner Party Seating Arrangements

A group of $n$ agents with numerical preferences for each other are to be assigned to the $n$ seats of a dining table. We study two natural topologies:~circular (cycle) tables and panel (path) tables. For a given seating arrangement, an agent's utility is the sum of their preference values towards their (at most two) direct neighbors. An arrangement is envy-free if no agent strictly prefers someone else's seat, and it is stable if no two agents strictly prefer each other's seats. Recently, it was shown that for both paths and cycles it is NP-hard to decide whether an envy-free arrangement exists, even for symmetric binary preferences. In contrast, we show that, if agents come from a bounded number of classes, the problem is solvable in polynomial time for arbitrarily-valued possibly asymmetric preferences, including outputting an arrangement if possible. We also give simpler proofs of the previous hardness results if preferences are allowed to be asymmetric. For stability, it is known that deciding the existence of stable arrangements is NP-hard for both topologies, but only if sufficiently-many numerical values are allowed. As it turns out, even constructing unstable instances can be challenging in certain cases, e.g., binary values. We completely characterize the existence of stable arrangements based on the number of distinct values in the preference matrix and the number of agent classes. We also ask the same question for non-negative values and give an almost-complete characterization, the most interesting outstanding case being that of paths with two-valued non-negative preferences, for which we experimentally find that stable arrangements always exist and prove it under the additional constraint that agents can only swap seats when sitting at most two positions away. We moreover give a polynomial algorithm for determining a stable arrangement assuming a bounded number of classes.

cs.GT↗

arXiv2023

DISCO-10M: A Large-Scale Music Dataset

Music datasets play a crucial role in advancing research in machine learning for music. However, existing music datasets suffer from limited size, accessibility, and lack of audio resources. To address these shortcomings, we present DISCO-10M, a novel and extensive music dataset that surpasses the largest previously available music dataset by an order of magnitude. To ensure high-quality data, we implement a multi-stage filtering process. This process incorporates similarities based on textual descriptions and audio embeddings. Moreover, we provide precomputed CLAP embeddings alongside DISCO-10M, facilitating direct application on various downstream tasks. These embeddings enable efficient exploration of machine learning applications on the provided data. With DISCO-10M, we aim to democratize and facilitate new research to help advance the development of novel machine learning models for music.

cs.SD↗

arXiv2023

What Determines the Price of NFTs?

In the evolving landscape of digital art, Non-Fungible Tokens (NFTs) have emerged as a groundbreaking platform, bridging the realms of art and technology. NFTs serve as the foundational framework that has revolutionized the market for digital art, enabling artists to showcase and monetize their creations in unprecedented ways. NFTs combine metadata stored on the blockchain with off-chain data, such as images, to create a novel form of digital ownership. It is not fully understood how these factors come together to determine NFT prices. In this study, we analyze both on-chain and off-chain data of NFT collections trading on OpenSea to understand what influences NFT pricing. Our results show that while text and image data of the NFTs can be used to explain price variations within collections, the extracted features do not generalize to new, unseen collections. Furthermore, we find that an NFT collection's trading volume often relates to its online presence, like social media followers and website traffic.

cs.CE↗

arXiv2023

The PoW Landscape in the Aftermath of The Merge

On 15th September 2022, The Merge marked the Ethereum network's transition from computation-hardness-based consensus (proof-of-work) to a committee-based consensus mechanism (proof-of-stake). As a result, all the specialized hardware and GPUs that were being used by miners ceased to be profitable in the main Ethereum network. Miners were then left with the decision of how to re-purpose their hardware. One such choice was to try and make a profit mining another existing PoW system. In this study, we explore this choice by analyzing the hashrate increase in the top PoW networks following the merge. Our findings reveal that the peak increase in hashrate to other PoW networks following The Merge represents an adoption of at least 41% of the hashrate that was present in Ethereum, with 12% remaining more than 5 months later. Though we measure a drastic decrease in profitability by almost an order of magnitude, the continued presence of miners halts claims that power consumption was instantly addressed by Ethereum's switch to PoS.

cs.CE↗

arXiv2023

Ethereum's Proposer-Builder Separation: Promises and Realities

With Ethereum's transition from Proof-of-Work to Proof-of-Stake in September 2022 came another paradigm shift, the Proposer-Builder Separation (PBS) scheme. PBS was introduced to decouple the roles of selecting and ordering transactions in a block (i.e., the builder), from those validating its contents and proposing the block to the network as the new head of the blockchain (i.e., the proposer). In this landscape, proposers are the validators in the Proof-of-Stake consensus protocol, while now relying on specialized block builders for creating blocks with the highest value for the proposer. Additionally, relays act as mediators between builders and proposers. We study PBS adoption and show that the current landscape exhibits significant centralization amongst the builders and relays. Further, we explore whether PBS effectively achieves its intended objectives of enabling hobbyist validators to maximize block profitability and preventing censorship. Our findings reveal that although PBS grants validators the opportunity to access optimized and competitive blocks, it tends to stimulate censorship rather than reduce it. Additionally, we demonstrate that relays do not consistently uphold their commitments and may prove unreliable. Specifically, proposers do not always receive the complete promised value, and the censorship or filtering capabilities pledged by relays exhibit significant gaps.

cs.DC↗

arXiv2023

Ethereum Proof-of-Stake Consensus Layer: Participation and Decentralization

In September 2022, Ethereum transitioned from Proof-of-Work (PoW) to Proof-of-Stake (PoS) during "the merge" - making it the largest PoS cryptocurrency in terms of market capitalization. With this work, we present a comprehensive measurement study of the current state of the Ethereum PoS consensus layer on the beacon chain. We perform a longitudinal study of the history of the beacon chain. Our work finds that all dips in network participation are caused by network upgrades, issues with major consensus clients, or issues with service operators controlling a large number of validators. Further, our longitudinal staking power decentralization analysis reveals that Ethereum PoS fairs similarly to its PoW counterpart in terms of decentralization and exhibits the immense impact of (liquid) staking services on staking power decentralization. Finally, we highlight the heightened security concerns in Ethereum PoS caused by high degrees of centralization.

cs.DC↗

arXiv2023

Fast Feedforward Networks

We break the linear link between the layer size and its inference cost by introducing the fast feedforward (FFF) architecture, a log-time alternative to feedforward networks. We demonstrate that FFFs are up to 220x faster than feedforward networks, up to 6x faster than mixture-of-experts networks, and exhibit better training properties than mixtures of experts thanks to noiseless conditional execution. Pushing FFFs to the limit, we show that they can use as little as 1% of layer neurons for inference in vision transformers while preserving 94.2% of predictive performance.

cs.LG↗

arXiv2023

DeFi Lending During The Merge

Lending protocols in decentralized finance enable the permissionless exchange of capital from lenders to borrowers without relying on a trusted third party for clearing or market-making. Interest rates are set by the supply and demand of capital according to a pre-defined function. In the lead-up to The Merge: Ethereum blockchain's transition from proof-of-work (PoW) to proof-of-stake (PoS), a fraction of the Ethereum ecosystem announced plans of continuing with a PoW-chain. Owners of ETH - whether their ETH was borrowed or not - would hold the native tokens on each chain. This development alarmed lending protocols. They feared spiking ETH borrowing rates would lead to mass liquidations which could undermine their viability. Thus, the decentralized autonomous organization running the protocols saw no alternative to intervention - restricting users' ability to borrow. We investigate the effects of the merge and the aforementioned intervention on the two biggest lending protocols on Ethereum: AAVE and Compound. Our analysis finds that borrowing rates were extremely volatile, jumping by two orders of magnitude, and borrowing at times reached 100% of the available funds. Despite this, no spike in mass liquidations or irretrievable loans materialized. Further, we are the first to quantify and analyze hard-fork-arbitrage, profiting from holding debt in the native blockchain token during a hard fork. We find that arbitrageurs transferred tokens to centralized exchanges which at the time were worth more than 13 Mio US$, money that was effectively extracted from the platforms' lenders.

q-fin.GN↗

arXiv2023

Diffusion Models for Graphs Benefit From Discrete State Spaces

Denoising diffusion probabilistic models and score-matching models have proven to be very powerful for generative tasks. While these approaches have also been applied to the generation of discrete graphs, they have, so far, relied on continuous Gaussian perturbations. Instead, in this work, we suggest using discrete noise for the forward Markov process. This ensures that in every intermediate step the graph remains discrete. Compared to the previous approach, our experimental results on four datasets and multiple architectures show that using a discrete noising process results in higher quality generated samples indicated with an average MMDs reduced by a factor of 1.5. Furthermore, the number of denoising steps is reduced from 1000 to 32 steps, leading to a 30 times faster sampling procedure.

cs.LG↗

arXiv2023

An Interpretable and Attention-based Method for Gaze Estimation Using Electroencephalography

Eye movements can reveal valuable insights into various aspects of human mental processes, physical well-being, and actions. Recently, several datasets have been made available that simultaneously record EEG activity and eye movements. This has triggered the development of various methods to predict gaze direction based on brain activity. However, most of these methods lack interpretability, which limits their technology acceptance. In this paper, we leverage a large data set of simultaneously measured Electroencephalography (EEG) and Eye tracking, proposing an interpretable model for gaze estimation from EEG data. More specifically, we present a novel attention-based deep learning framework for EEG signal analysis, which allows the network to focus on the most relevant information in the signal and discard problematic channels. Additionally, we provide a comprehensive evaluation of the presented framework, demonstrating its superiority over current methods in terms of accuracy and robustness. Finally, the study presents visualizations that explain the results of the analysis and highlights the potential of attention mechanism for improving the efficiency and effectiveness of EEG data analysis in a variety of applications.

eess.SP↗

arXiv2023

Graphtester: Exploring Theoretical Boundaries of GNNs on Graph Datasets

Graph Neural Networks (GNNs) have emerged as a powerful tool for learning from graph-structured data. However, even state-of-the-art architectures have limitations on what structures they can distinguish, imposing theoretical limits on what the networks can achieve on different datasets. In this paper, we provide a new tool called Graphtester for a comprehensive analysis of the theoretical capabilities of GNNs for various datasets, tasks, and scores. We use Graphtester to analyze over 40 different graph datasets, determining upper bounds on the performance of various GNNs based on the number of layers. Further, we show that the tool can also be used for Graph Transformers using positional node encodings, thereby expanding its scope. Finally, we demonstrate that features generated by Graphtester can be used for practical applications such as Graph Transformers, and provide a synthetic dataset to benchmark node and edge features, such as positional encodings. The package is freely available at the following URL: https://github.com/meakbiyik/graphtester.

cs.LG↗

arXiv2023

Abstract Visual Reasoning Enabled by Language

While artificial intelligence (AI) models have achieved human or even superhuman performance in many well-defined applications, they still struggle to show signs of broad and flexible intelligence. The Abstraction and Reasoning Corpus (ARC), a visual intelligence benchmark introduced by François Chollet, aims to assess how close AI systems are to human-like cognitive abilities. Most current approaches rely on carefully handcrafted domain-specific program searches to brute-force solutions for the tasks present in ARC. In this work, we propose a general learning-based framework for solving ARC. It is centered on transforming tasks from the vision to the language domain. This composition of language and vision allows for pre-trained models to be leveraged at each stage, enabling a shift from handcrafted priors towards the learned priors of the models. While not yet beating state-of-the-art models on ARC, we demonstrate the potential of our approach, for instance, by solving some ARC tasks that have not been solved previously.

cs.AI↗

arXiv2023

Siamese SIREN: Audio Compression with Implicit Neural Representations

Implicit Neural Representations (INRs) have emerged as a promising method for representing diverse data modalities, including 3D shapes, images, and audio. While recent research has demonstrated successful applications of INRs in image and 3D shape compression, their potential for audio compression remains largely unexplored. Motivated by this, we present a preliminary investigation into the use of INRs for audio compression. Our study introduces Siamese SIREN, a novel approach based on the popular SIREN architecture. Our experimental results indicate that Siamese SIREN achieves superior audio reconstruction fidelity while utilizing fewer network parameters compared to previous INR architectures.

cs.SD↗

arXiv2023

Short Squeeze in DeFi Lending Market: Decentralization in Jeopardy?

Anxiety levels in the Aave community spiked in November 2022 as Avi Eisenberg performed an attack on Aave. Eisenberg attempted to short the CRV token by using funds borrowed on the protocol to artificially deflate the value of CRV. While the attack was ultimately unsuccessful, it left the Aave community scared and even raised question marks regarding the feasibility of large lending platforms under decentralized governance. In this work, we analyze Avi Eisenberg's actions and show how he was able to artificially lower the price of CRV by selling large quantities of borrowed CRV for stablecoins on both decentralized and centralized exchanges. Despite the failure of his attack, it still led to irretrievable debt worth more than 1.5 Mio USD at the time and, thereby, quadrupled the protocol's irretrievable debt. Furthermore, we highlight that his attack was enabled by the vast proportion of CRV available to borrow as well as Aave's lending protocol design hindering rapid intervention. We stress Eisenberg's attack exposes a predicament of large DeFi lending protocols: limit the scope or compromise on 'decentralization'.

cs.CR↗

arXiv2023

Examining the Emergence of Deductive Reasoning in Generative Language Models

We conduct a preliminary inquiry into the ability of generative transformer models to deductively reason from premises provided. We observe notable differences in the performance of models coming from different training setups and find that the deductive reasoning ability increases with scale. Further, we discover that the performance generally does not decrease with the length of the deductive chain needed to reach the conclusion, with the exception of OpenAI GPT-3 and GPT-3.5 models. Our study considers a wide variety of transformer-decoder models, ranging from 117 million to 175 billion parameters in size.

cs.CL↗

arXiv2023

Cascaded Beam Search: Plug-and-Play Terminology-Forcing For Neural Machine Translation

This paper presents a plug-and-play approach for translation with terminology constraints. Terminology constraints are an important aspect of many modern translation pipelines. In both specialized domains and newly emerging domains (such as the COVID-19 pandemic), accurate translation of technical terms is crucial. Recent approaches often train models to copy terminologies from the input into the output sentence by feeding the target terminology along with the input. But this requires expensive training whenever the underlying language model is changed or the system should specialize to a new domain. We propose Cascade Beam Search, a plug-and-play terminology-forcing approach that requires no training. Cascade Beam Search has two parts: 1) logit manipulation to increase the probability of target terminologies and 2) a cascading beam setup based on grid beam search, where beams are grouped by the number of terminologies they contain. We evaluate the performance of our approach by competing against the top submissions of the WMT21 terminology translation task. Our plug-and-play approach performs on par with the winning submissions without using a domain-specific language model and with no additional training.

cs.CL↗

arXiv2023

Divide and Scale: Formalization and Roadmap to Robust Sharding

Sharding distributed ledgers is a promising on-chain solution for scaling blockchains but lacks formal grounds, nurturing skepticism on whether such complex systems can scale blockchains securely. We fill this gap by introducing the first formal framework as well as a roadmap to robust sharding. In particular, we first define the properties sharded distributed ledgers should fulfill. We build upon and extend the Bitcoin backbone protocol by defining consistency and scalability. Consistency encompasses the need for atomic execution of cross-shard transactions to preserve safety, whereas scalability encapsulates the speedup a sharded system can gain in comparison to a non-sharded system. Using our model, we explore the limitations of sharding. We show that a sharded ledger with $n$ participants cannot scale under a fully adaptive adversary, but it can scale up to $m$ shards where $n=c'm\log m$, under an epoch-adaptive adversary; the constant $c'$ encompasses the trade-off between security and scalability. This is possible only if the sharded ledgers create succinct proofs of the valid state updates at every epoch. We leverage our results to identify the sufficient components for robust sharding, which we incorporate in a protocol abstraction termed Divide & Scale. To demonstrate the power of our framework, we analyze the most prominent sharded blockchains (Elastico, Monoxide, OmniLedger, RapidChain) and pinpoint where they fail to meet the desired properties.

cs.DC↗

arXiv2023

Print Your Money: Cash-Like Experiences with Digital Money

The use of digital money has become increasingly popular, but it comes with certain drawbacks. For instance, it can be challenging to make payments during power outages or internet failures. Additionally, some groups may find it difficult to use digital money. To address these concerns, we propose a design for a central bank digital currency (CBDC) similar to physical cash but also integrates with digital payment systems. This would enable users to access digital money without needing a third party. Our design also addresses technical and security concerns by implementing a trust-level model and ensuring that the system meets users' security needs. Ultimately, our design has the potential to replace physical banknotes and coins.

cs.HC↗