Positional Encoding Helps Recurrent Neural Networks Handle a Large Vocabulary

This study discusses the effects of positional encoding on recurrent neural networks (RNNs) utilizing synthetic benchmarks. Positional encoding "time-stamps" data points in time series and complements the capabilities of Transformer neural networks, which lack an inherent mechanism for representing the data order. By contrast, RNNs can encode the temporal information of data points on their own, rendering their use of positional encoding seemingly "redundant". Nonetheless, empirical investigations reveal the effectiveness of positional encoding even when coupled with RNNs, specifically for handling a large vocabulary that yields diverse observations. These findings pave the way for a new line of research on RNNs, concerning the combination of input-driven and autonomous time representation. Additionally, biological implications of the computational/simulational results are discussed, in the light of the affinity between the sinusoidal implementation of positional encoding and neural oscillations in biological brains.

HoughToRadon Transform: New Neural Network Layer for Features Improvement in Projection Space

In this paper, we introduce HoughToRadon Transform layer, a novel layer designed to improve the speed of neural networks incorporated with Hough Transform to solve semantic image segmentation problems. By placing it after a Hough Transform layer, "inner" convolutions receive modified feature maps with new beneficial properties, such as a smaller area of processed images and parameter space linearity by angle and shift. These properties were not presented in Hough Transform alone. Furthermore, HoughToRadon Transform layer allows us to adjust the size of intermediate feature maps using two new parameters, thus allowing us to balance the speed and quality of the resulting neural network. Our experiments on the open MIDV-500 dataset show that this new approach leads to time savings in document segmentation tasks and achieves state-of-the-art 97.7% accuracy, outperforming HoughEncoder with larger computational complexity.

Graph-enhanced Large Language Models in Asynchronous Plan Reasoning

Reasoning about asynchronous plans is challenging since it requires sequential and parallel planning to optimize time costs. Can large language models (LLMs) succeed at this task? Here, we present the first large-scale study investigating this question. We find that a representative set of closed and open-source LLMs, including GPT-4 and LLaMA-2, behave poorly when not supplied with illustrations about the task-solving process in our benchmark AsyncHow. We propose a novel technique called Plan Like a Graph (PLaG) that combines graphs with natural language prompts and achieves state-of-the-art results. We show that although PLaG can boost model performance, LLMs still suffer from drastic degradation when task complexity increases, highlighting the limits of utilizing LLMs for simulating digital devices. We see our study as an exciting step towards using LLMs as efficient autonomous agents.

The last Dance : Robust backdoor attack via diffusion models and bayesian approach

Diffusion models are state-of-the-art deep learning generative models that are trained on the principle of learning forward and backward diffusion processes via the progressive addition of noise and denoising. In this paper, we seek to trick audio-based DNN models, such as those in the Hugging Face framework, for example, those that focus on audio, in particular transformer-based artificial intelligence models, which are powerful machine learning models that save time and deliver faster, more efficient results. We demonstrate the feasibility of backdoor attacks (called `BacKBayDiffMod`) on audio transformers derived from Hugging Face, a popular framework in the world of artificial intelligence (AI) research. The backdoor attack developed in this paper is based on poisoning the model's training data by incorporating backdoor diffusion sampling and a Bayesian approach to the distribution of poisoned data.

Learning Time-aware Graph Structures for Spatially Correlated Time Series Forecasting

Spatio-temporal forecasting of future values of spatially correlated time series is important across many cyber-physical systems (CPS). Recent studies offer evidence that the use of graph neural networks to capture latent correlations between time series holds a potential for enhanced forecasting. However, most existing methods rely on pre-defined or self-learning graphs, which are either static or unintentionally dynamic, and thus cannot model the time-varying correlations that exhibit trends and periodicities caused by the regularity of the underlying processes in CPS. To tackle such limitation, we propose Time-aware Graph Structure Learning (TagSL), which extracts time-aware correlations among time series by measuring the interaction of node and time representations in high-dimensional spaces. Notably, we introduce time discrepancy learning that utilizes contrastive learning with distance-based regularization terms to constrain learned spatial correlations to a trend sequence. Additionally, we propose a periodic discriminant function to enable the capture of periodic changes from the state of nodes. Next, we present a Graph Convolution-based Gated Recurrent Unit (GCGRU) that jointly captures spatial and temporal dependencies while learning time-aware and node-specific patterns. Finally, we introduce a unified framework named Time-aware Graph Convolutional Recurrent Network (TGCRN), combining TagSL, and GCGRU in an encoder-decoder architecture for multi-step spatio-temporal forecasting. We report on experiments with TGCRN and popular existing approaches on five real-world datasets, thus providing evidence that TGCRN is capable of advancing the state-of-the-art. We also cover a detailed ablation study and visualization analysis, offering detailed insight into the effectiveness of time-aware structure learning.

Locally-Adaptive Quantization for Streaming Vector Search

Retrieving the most similar vector embeddings to a given query among a massive collection of vectors has long been a key component of countless real-world applications. The recently introduced Retrieval-Augmented Generation is one of the most prominent examples. For many of these applications, the database evolves over time by inserting new data and removing outdated data. In these cases, the retrieval problem is known as streaming similarity search. While Locally-Adaptive Vector Quantization (LVQ), a highly efficient vector compression method, yields state-of-the-art search performance for non-evolving databases, its usefulness in the streaming setting has not been yet established. In this work, we study LVQ in streaming similarity search. In support of our evaluation, we introduce two improvements of LVQ: Turbo LVQ and multi-means LVQ that boost its search performance by up to 28% and 27%, respectively. Our studies show that LVQ and its new variants enable blazing fast vector search, outperforming its closest competitor by up to 9.4x for identically distributed data and by up to 8.8x under the challenging scenario of data distribution shifts (i.e., where the statistical distribution of the data changes over time). We release our contributions as part of Scalable Vector Search, an open-source library for high-performance similarity search.

Low-rank Attention Side-Tuning for Parameter-Efficient Fine-Tuning

In finetuning a large pretrained model to downstream tasks, parameter-efficient fine-tuning (PEFT) methods can effectively finetune pretrained models with few trainable parameters, but suffer from high GPU memory consumption and slow training speed. Because learnable parameters from these methods are entangled with the pretrained model, gradients related to the frozen pretrained model's parameters have to be computed and stored during finetuning. We propose Low-rank Attention Side-Tuning (LAST), which disentangles the trainable module from the pretrained model by freezing not only parameters but also outputs of the pretrained network. LAST trains a side-network composed of only low-rank self-attention modules. By viewing the pretrained model as a frozen feature extractor, the side-network takes intermediate output from the pretrained model and focus on learning task-specific knowledge. We also show that LAST can be highly parallel across multiple optimization objectives, making it very efficient in downstream task adaptation, for example, in finding optimal hyperparameters. LAST outperforms previous state-of-the-art methods on VTAB-1K and other visual adaptation tasks with roughly only 30\% of GPU memory footprint and 60\% of training time compared to existing PEFT methods, but achieves significantly higher accuracy.

Prediction Horizon Requirements for Automated Driving: Optimizing Safety, Comfort, and Efficiency

Predicting the movement of other road users is beneficial for improving automated vehicle (AV) performance. However, the relationship between the time horizon associated with these predictions and AV performance remains unclear. Despite the existence of numerous trajectory prediction algorithms, no studies have been conducted on how varying prediction lengths affect AV safety and other vehicle performance metrics, resulting in undefined horizon requirements for prediction methods. Our study addresses this gap by examining the effects of different prediction horizons on AV performance, focusing on safety, comfort, and efficiency. Through multiple experiments using a state-of-the-art, risk-based predictive trajectory planner, we simulated predictions with horizons up to 20 seconds. Based on our simulations, we propose a framework for specifying the minimum required and optimal prediction horizons based on specific AV performance criteria and application needs. Our results indicate that a horizon of 1.6 seconds is required to prevent collisions with crossing pedestrians, horizons of 7-8 seconds yield the best efficiency, and horizons up to 15 seconds improve passenger comfort. We conclude that prediction horizon requirements are application-dependent, and recommend aiming for a prediction horizon of 11.8 seconds as a general guideline for applications involving crossing pedestrians.

Are we making much progress? Revisiting chemical reaction yield prediction from an imbalanced regression perspective

The yield of a chemical reaction quantifies the percentage of the target product formed in relation to the reactants consumed during the chemical reaction. Accurate yield prediction can guide chemists toward selecting high-yield reactions during synthesis planning, offering valuable insights before dedicating time and resources to wet lab experiments. While recent advancements in yield prediction have led to overall performance improvement across the entire yield range, an open challenge remains in enhancing predictions for high-yield reactions, which are of greater concern to chemists. In this paper, we argue that the performance gap in high-yield predictions results from the imbalanced distribution of real-world data skewed towards low-yield reactions, often due to unreacted starting materials and inherent ambiguities in the reaction processes. Despite this data imbalance, existing yield prediction methods continue to treat different yield ranges equally, assuming a balanced training distribution. Through extensive experiments on three real-world yield prediction datasets, we emphasize the urgent need to reframe reaction yield prediction as an imbalanced regression problem. Finally, we demonstrate that incorporating simple cost-sensitive re-weighting methods can significantly enhance the performance of yield prediction models on underrepresented high-yield regions.

QuEST: Low-bit Diffusion Model Quantization via Efficient Selective Finetuning

Diffusion models have achieved remarkable success in image generation tasks, yet their practical deployment is restrained by the high memory and time consumption. While quantization paves a way for diffusion model compression and acceleration, existing methods totally fail when the models are quantized to low-bits. In this paper, we unravel three properties in quantized diffusion models that compromise the efficacy of current methods: imbalanced activation distributions, imprecise temporal information, and vulnerability to perturbations of specific modules. To alleviate the intensified low-bit quantization difficulty stemming from the distribution imbalance, we propose finetuning the quantized model to better adapt to the activation distribution. Building on this idea, we identify two critical types of quantized layers: those holding vital temporal information and those sensitive to reduced bit-width, and finetune them to mitigate performance degradation with efficiency. We empirically verify that our approach modifies the activation distribution and provides meaningful temporal information, facilitating easier and more accurate quantization. Our method is evaluated over three high-resolution image generation tasks and achieves state-of-the-art performance under various bit-width settings, as well as being the first method to generate readable images on full 4-bit (i.e. W4A4) Stable Diffusion.