Rana Ali Amjad

Rana Ali Amjad, Clemens Blöchl, Bernhard C. Geiger

This paper proposes an information-theoretic cost function for aggregating a Markov chain via a (possibly stochastic) mapping. The cost function is motivated by two objectives: 1) The process obtained by observing the Markov chain through the mapping should be close to a Markov chain, and 2) the aggregated Markov chain should retain as much of the temporal dependence structure of the original Markov chain as possible. We discuss properties of this parameterized cost function and show that it contains the cost functions previously proposed by Deng et al., Xu et al., and Geiger et al. as special cases. We moreover discuss these special cases providing a better understanding and highlighting potential shortcomings: For example, the cost function proposed by Geiger et al. is tightly connected to approximate probabilistic bisimulation, but leads to trivial solutions if optimized without regularization. We furthermore propose a simple heuristic to optimize our cost function for deterministic aggregations and illustrate its performance on a set of synthetic examples.

Zhining Liu, Rana Ali Amjad, Ravinarayana Adkathimar et al.

Providing Language Models (LMs) with relevant evidence in the context (either via retrieval or user-provided) can significantly improve their ability to provide better-grounded responses. However, recent studies have found that LMs often struggle to fully comprehend and utilize key evidence from the context, especially when it contains noise and irrelevant information, an issue common in real-world scenarios. To address this, we propose SelfElicit, an inference-time approach that helps LMs focus on key contextual evidence through self-guided explicit highlighting. By leveraging the inherent evidence-finding capabilities of LMs using the attention scores of deeper layers, our method automatically identifies and emphasizes key evidence within the input context, facilitating more accurate and grounded responses without additional training or iterative prompting. We demonstrate that SelfElicit brings consistent and significant improvement on multiple evidence-based QA tasks for various LM families while maintaining computational efficiency. Our code and documentation are available at https://github.com/ZhiningLiu1998/SelfElicit.

Sindhu Tipirneni, Ravinarayana Adkathimar, Nurendra Choudhary et al.

Despite the remarkable success of Large Language Models (LLMs) in text understanding and generation, their potential for text clustering tasks remains underexplored. We observed that powerful closed-source LLMs provide good quality clusterings of entity sets but are not scalable due to the massive compute power required and the associated costs. Thus, we propose CACTUS (Context-Aware ClusTering with aUgmented triplet losS), a systematic approach that leverages open-source LLMs for efficient and effective supervised clustering of entity subsets, particularly focusing on text-based entities. Existing text clustering methods fail to effectively capture the context provided by the entity subset. Moreover, though there are several language modeling based approaches for clustering, very few are designed for the task of supervised clustering. This paper introduces a novel approach towards clustering entity subsets using LLMs by capturing context via a scalable inter-entity attention mechanism. We propose a novel augmented triplet loss function tailored for supervised clustering, which addresses the inherent challenges of directly applying the triplet loss to this problem. Furthermore, we introduce a self-supervised clustering task based on text augmentation techniques to improve the generalization of our model. For evaluation, we collect ground truth clusterings from a closed-source LLM and transfer this knowledge to an open-source LLM under the supervised clustering framework, allowing a faster and cheaper open-source model to perform the same task. Experiments on various e-commerce query and product clustering datasets demonstrate that our proposed approach significantly outperforms existing unsupervised and supervised baselines under various external clustering evaluation metrics.

Kumar Pratik, Rana Ali Amjad, Arash Behboodi et al.

We propose Hypernetwork Kalman Filter (HKF) for tracking applications with multiple different dynamics. The HKF combines generalization power of Kalman filters with expressive power of neural networks. Instead of keeping a bank of Kalman filters and choosing one based on approximating the actual dynamics, HKF adapts itself to each dynamics based on the observed sequence. Through extensive experiments on CDL-B channel model, we show that the HKF can be used for tracking the channel over a wide range of Doppler values, matching Kalman filter performance with genie Doppler information. At high Doppler values, it achieves around 2dB gain over genie Kalman filter. The HKF generalizes well to unseen Doppler, SNR values and pilot patterns unlike LSTM, which suffers from severe performance degradation.

Markus Nagel, Marios Fournarakis, Rana Ali Amjad et al.

While neural networks have advanced the frontiers in many applications, they often come at a high computational cost. Reducing the power and latency of neural network inference is key if we want to integrate modern networks into edge devices with strict power and compute requirements. Neural network quantization is one of the most effective ways of achieving these savings but the additional noise it induces can lead to accuracy degradation. In this white paper, we introduce state-of-the-art algorithms for mitigating the impact of quantization noise on the network's performance while maintaining low-bit weights and activations. We start with a hardware motivated introduction to quantization and then consider two main classes of algorithms: Post-Training Quantization (PTQ) and Quantization-Aware-Training (QAT). PTQ requires no re-training or labelled data and is thus a lightweight push-button approach to quantization. In most cases, PTQ is sufficient for achieving 8-bit quantization with close to floating-point accuracy. QAT requires fine-tuning and access to labeled training data but enables lower bit quantization with competitive results. For both solutions, we provide tested pipelines based on existing literature and extensive experimentation that lead to state-of-the-art performance for common deep learning models and tasks.

Mart van Baalen, Christos Louizos, Markus Nagel et al.

We introduce Bayesian Bits, a practical method for joint mixed precision quantization and pruning through gradient based optimization. Bayesian Bits employs a novel decomposition of the quantization operation, which sequentially considers doubling the bit width. At each new bit width, the residual error between the full precision value and the previously rounded value is quantized. We then decide whether or not to add this quantized residual error for a higher effective bit width and lower quantization noise. By starting with a power-of-two bit width, this decomposition will always produce hardware-friendly configurations, and through an additional 0-bit option, serves as a unified view of pruning and quantization. Bayesian Bits then introduces learnable stochastic gates, which collectively control the bit width of the given tensor. As a result, we can obtain low bit solutions by performing approximate inference over the gates, with prior distributions that encourage most of them to be switched off. We experimentally validate our proposed method on several benchmark datasets and show that we can learn pruned, mixed precision networks that provide a better trade-off between accuracy and efficiency than their static bit width equivalents.

Markus Nagel, Rana Ali Amjad, Mart van Baalen et al.

When quantizing neural networks, assigning each floating-point weight to its nearest fixed-point value is the predominant approach. We find that, perhaps surprisingly, this is not the best we can do. In this paper, we propose AdaRound, a better weight-rounding mechanism for post-training quantization that adapts to the data and the task loss. AdaRound is fast, does not require fine-tuning of the network, and only uses a small amount of unlabelled data. We start by theoretically analyzing the rounding problem for a pre-trained neural network. By approximating the task loss with a Taylor series expansion, the rounding task is posed as a quadratic unconstrained binary optimization problem. We simplify this to a layer-wise local loss and propose to optimize this loss with a soft relaxation. AdaRound not only outperforms rounding-to-nearest by a significant margin but also establishes a new state-of-the-art for post-training quantization on several networks and tasks. Without fine-tuning, we can quantize the weights of Resnet18 and Resnet50 to 4 bits while staying within an accuracy loss of 1%.

Rana Ali Amjad, Bernhard C. Geiger

In this draft, which reports on work in progress, we 1) adapt the information bottleneck functional by replacing the compression term by class-conditional compression, 2) relax this functional using a variational bound related to class-conditional disentanglement, 3) consider this functional as a training objective for stochastic neural networks, and 4) show that the latent representations are learned such that they can be used in a naive Bayes classifier. We continue by suggesting a series of experiments along the lines of Nonlinear In-formation Bottleneck [Kolchinsky et al., 2018], Deep Variational Information Bottleneck [Alemi et al., 2017], and Information Dropout [Achille and Soatto, 2018]. We furthermore suggest a neural network where the decoder architecture is a parameterized naive Bayes decoder.

Rana Ali Amjad, Kairen Liu, Bernhard C. Geiger

In this work, we investigate the use of three information-theoretic quantities -- entropy, mutual information with the class variable, and a class selectivity measure based on Kullback-Leibler divergence -- to understand and study the behavior of already trained fully-connected feed-forward neural networks. We analyze the connection between these information-theoretic quantities and classification performance on the test set by cumulatively ablating neurons in networks trained on MNIST, FashionMNIST, and CIFAR-10. Our results parallel those recently published by Morcos et al., indicating that class selectivity is not a good indicator for classification performance. However, looking at individual layers separately, both mutual information and class selectivity are positively correlated with classification performance, at least for networks with ReLU activation functions. We provide explanations for this phenomenon and conclude that it is ill-advised to compare the proposed information-theoretic quantities across layers. Furthermore, we show that cumulative ablation of neurons with ascending or descending information-theoretic quantities can be used to formulate hypotheses regarding the joint behavior of multiple neurons, such as redundancy and synergy, with comparably low computational cost. We also draw connections to the information bottleneck theory for neural networks.

Rayyan Ahmad Khan, Rana Ali Amjad, Martin Kleinsteuber

We propose a new clustering algorithm, Extended Affinity Propagation, based on pairwise similarities. Extended Affinity Propagation is developed by modifying Affinity Propagation such that the desirable features of Affinity Propagation, e.g., exemplars, reasonable computational complexity and no need to specify number of clusters, are preserved while the shortcomings, e.g., the lack of global structure discovery, that limit the applicability of Affinity Propagation are overcome. Extended Affinity Propagation succeeds not only in achieving this goal but can also provide various additional insights into the internal structure of the individual clusters, e.g., refined confidence values, relative cluster densities and local cluster strength in different regions of a cluster, which are valuable for an analyst. We briefly discuss how these insights can help in easily tuning the hyperparameters. We also illustrate these desirable features and the performance of Extended Affinity Propagation on various synthetic and real world datasets.

Rana Ali Amjad, Bernhard C. Geiger

In this theory paper, we investigate training deep neural networks (DNNs) for classification via minimizing the information bottleneck (IB) functional. We show that the resulting optimization problem suffers from two severe issues: First, for deterministic DNNs, either the IB functional is infinite for almost all values of network parameters, making the optimization problem ill-posed, or it is piecewise constant, hence not admitting gradient-based optimization methods. Second, the invariance of the IB functional under bijections prevents it from capturing properties of the learned representation that are desirable for classification, such as robustness and simplicity. We argue that these issues are partly resolved for stochastic DNNs, DNNs that include a (hard or soft) decision rule, or by replacing the IB functional with related, but more well-behaved cost functions. We conclude that recent successes reported about training DNNs using the IB framework must be attributed to such solutions. As a side effect, our results indicate limitations of the IB framework for the analysis of DNNs. We also note that rather than trying to repair the inherent problems in the IB functional, a better approach may be to design regularizers on latent representation enforcing the desired properties directly.

Clemens Bloechl, Rana Ali Amjad, Bernhard C. Geiger

We present an information-theoretic cost function for co-clustering, i.e., for simultaneous clustering of two sets based on similarities between their elements. By constructing a simple random walk on the corresponding bipartite graph, our cost function is derived from a recently proposed generalized framework for information-theoretic Markov chain aggregation. The goal of our cost function is to minimize relevant information loss, hence it connects to the information bottleneck formalism. Moreover, via the connection to Markov aggregation, our cost function is not ad hoc, but inherits its justification from the operational qualities associated with the corresponding Markov aggregation problem. We furthermore show that, for appropriate parameter settings, our cost function is identical to well-known approaches from the literature, such as Information-Theoretic Co-Clustering of Dhillon et al. Hence, understanding the influence of this parameter admits a deeper understanding of the relationship between previously proposed information-theoretic cost functions. We highlight some strengths and weaknesses of the cost function for different parameters. We also illustrate the performance of our cost function, optimized with a simple sequential heuristic, on several synthetic and real-world data sets, including the Newsgroup20 and the MovieLens100k data sets.

Bernhard C. Geiger, Rana Ali Amjad

In this paper, we investigate mutual information as a cost function for clustering, and show in which cases hard, i.e., deterministic, clusters are optimal. Using convexity properties of mutual information, we show that certain formulations of the information bottleneck problem are solved by hard clusters. Similarly, hard clusters are optimal for the information-theoretic co-clustering problem that deals with simultaneous clustering of two dependent data sets. If both data sets have to be clustered using the same cluster assignment, hard clusters are not optimal in general. We point at interesting and practically relevant special cases of this so-called pairwise clustering problem, for which we can either prove or have evidence that hard clusters are optimal. Our results thus show that one can relax the otherwise combinatorial hard clustering problem to a real-valued optimization problem with the same global optimum.