arXiv subjects

Kannan Ramchandran

Publications and source records attributed to Kannan Ramchandran.

At least 55 records · Page 3Linked to original sources

arXiv2022

Decentralized Competing Bandits in Non-Stationary Matching Markets

Understanding complex dynamics of two-sided online matching markets, where the demand-side agents compete to match with the supply-side (arms), has recently received substantial interest. To that end, in this paper, we introduce the framework of decentralized two-sided matching market under non stationary (dynamic) environments. We adhere to the serial dictatorship setting, where the demand-side agents have unknown and different preferences over the supply-side (arms), but the arms have fixed and known preference over the agents. We propose and analyze a decentralized and asynchronous learning algorithm, namely Decentralized Non-stationary Competing Bandits (\texttt{DNCB}), where the agents play (restrictive) successive elimination type learning algorithms to learn their preference over the arms. The complexity in understanding such a system stems from the fact that the competing bandits choose their actions in an asynchronous fashion, and the lower ranked agents only get to learn from a set of arms, not \emph{dominated} by the higher ranked agents, which leads to \emph{forced exploration}. With carefully defined complexity parameters, we characterize this \emph{forced exploration} and obtain sub-linear (logarithmic) regret of \texttt{DNCB}. Furthermore, we validate our theoretical findings via experiments.

stat.ML↗

arXiv2022

Adaptive Clustering and Personalization in Multi-Agent Stochastic Linear Bandits

We consider the problem of minimizing regret in an $N$ agent heterogeneous stochastic linear bandits framework, where the agents (users) are similar but not all identical. We model user heterogeneity using two popularly used ideas in practice; (i) A clustering framework where users are partitioned into groups with users in the same group being identical to each other, but different across groups, and (ii) a personalization framework where no two users are necessarily identical, but a user's parameters are close to that of the population average. In the clustered users' setup, we propose a novel algorithm, based on successive refinement of cluster identities and regret minimization. We show that, for any agent, the regret scales as $\mathcal{O}(\sqrt{T/N})$, if the agent is in a `well separated' cluster, or scales as $\mathcal{O}(T^{\frac{1}{2} + \varepsilon}/(N)^{\frac{1}{2} -\varepsilon})$ if its cluster is not well separated, where $\varepsilon$ is positive and arbitrarily close to $0$. Our algorithm is adaptive to the cluster separation, and is parameter free -- it does not need to know the number of clusters, separation and cluster size, yet the regret guarantee adapts to the inherent complexity. In the personalization framework, we introduce a natural algorithm where, the personal bandit instances are initialized with the estimates of the global average model. We show that, an agent $i$ whose parameter deviates from the population average by $ε_i$, attains a regret scaling of $\widetilde{O}(ε_i\sqrt{T})$. This demonstrates that if the user representations are close (small $ε_i)$, the resulting regret is low, and vice-versa. The results are empirically validated and we observe superior performance of our adaptive algorithms over non-adaptive baselines.

stat.ML↗

arXiv2022

Utility-based Resource Allocation and Pricing for Serverless Computing

Serverless computing platforms currently rely on basic pricing schemes that are static and do not reflect customer feedback. This leads to significant inefficiencies from a total utility perspective. As one of the fastest-growing cloud services, serverless computing provides an opportunity to better serve both users and providers through the incorporation of market-based strategies for pricing and resource allocation. With the help of utility functions to model the delay-sensitivity of customers, we propose a novel scheduler to allocate resources for serverless computing. The resulting resource allocation scheme is optimal in the sense that it maximizes the aggregate utility of all users across the system, thus maximizing social welfare. Our approach gives rise to a natural dynamic pricing scheme that is obtained by solving an optimization problem in its dual form. We further develop feedback mechanisms that allow the cloud provider to converge to optimal resource allocation, even when the users' utilities are private and unknown to the service provider. Simulations show that our approach can track market demand and achieve significantly higher social welfare (or, equivalently, cost savings for customers) compared to existing schemes.

cs.DC↗

arXiv2021

Escaping Saddle Points in Distributed Newton's Method with Communication Efficiency and Byzantine Resilience

The problem of saddle-point avoidance for non-convex optimization is quite challenging in large scale distributed learning frameworks, such as Federated Learning, especially in the presence of Byzantine workers. The celebrated cubic-regularized Newton method of \cite{nest} is one of the most elegant ways to avoid saddle-points in the standard centralized (non-distributed) setup. In this paper, we extend the cubic-regularized Newton method to a distributed framework and simultaneously address several practical challenges like communication bottleneck and Byzantine attacks. Note that the issue of saddle-point avoidance becomes more crucial in the presence of Byzantine machines since rogue machines may create \emph{fake local minima} near the saddle-points of the loss function, also known as the saddle-point attack. Being a second order algorithm, our iteration complexity is much lower than the first order counterparts. Furthermore we use compression (or sparsification) techniques like $δ$-approximate compression for communication efficiency. We obtain theoretical guarantees for our proposed scheme under several settings including approximate (sub-sampled) gradients and Hessians. Moreover, we validate our theoretical findings with experiments using standard datasets and several types of Byzantine attacks, and obtain an improvement of $25\%$ with respect to first order methods in iteration complexity.

cs.DC↗

arXiv2021

Taxonomizing local versus global structure in neural network loss landscapes

Viewing neural network models in terms of their loss landscapes has a long history in the statistical mechanics approach to learning, and in recent years it has received attention within machine learning proper. Among other things, local metrics (such as the smoothness of the loss landscape) have been shown to correlate with global properties of the model (such as good generalization performance). Here, we perform a detailed empirical analysis of the loss landscape structure of thousands of neural network models, systematically varying learning tasks, model architectures, and/or quantity/quality of data. By considering a range of metrics that attempt to capture different aspects of the loss landscape, we demonstrate that the best test accuracy is obtained when: the loss landscape is globally well-connected; ensembles of trained models are more similar to each other; and models converge to locally smooth regions. We also show that globally poorly-connected landscapes can arise when models are small or when they are trained to lower quality data; and that, if the loss landscape is globally poorly-connected, then training to zero loss can actually lead to worse test accuracy. Our detailed empirical results shed light on phases of learning (and consequent double descent behavior), fundamental versus incidental determinants of good generalization, the role of load-like and temperature-like parameters in the learning process, different influences on the loss landscape from model and data, and the relationships between local and global metrics, all topics of recent interest.

cs.LG↗

arXiv2021

Model Selection for Generic Reinforcement Learning

We address the problem of model selection for the finite horizon episodic Reinforcement Learning (RL) problem where the transition kernel $P^*$ belongs to a family of models $\mathcal{P}^*$ with finite metric entropy. In the model selection framework, instead of $\mathcal{P}^*$, we are given $M$ nested families of transition kernels $\cP_1 \subset \cP_2 \subset \ldots \subset \cP_M$. We propose and analyze a novel algorithm, namely \emph{Adaptive Reinforcement Learning (General)} (\texttt{ARL-GEN}) that adapts to the smallest such family where the true transition kernel $P^*$ lies. \texttt{ARL-GEN} uses the Upper Confidence Reinforcement Learning (\texttt{UCRL}) algorithm with value targeted regression as a blackbox and puts a model selection module at the beginning of each epoch. Under a mild separability assumption on the model classes, we show that \texttt{ARL-GEN} obtains a regret of $\Tilde{\mathcal{O}}(d_{\mathcal{E}}^*H^2+\sqrt{d_{\mathcal{E}}^* \mathbb{M}^* H^2 T})$, with high probability, where $H$ is the horizon length, $T$ is the total number of steps, $d_{\mathcal{E}}^*$ is the Eluder dimension and $\mathbb{M}^*$ is the metric entropy corresponding to $\mathcal{P}^*$. Note that this regret scaling matches that of an oracle that knows $\mathcal{P}^*$ in advance. We show that the cost of model selection for \texttt{ARL-GEN} is an additive term in the regret having a weak dependence on $T$. Subsequently, we remove the separability assumption and consider the setup of linear mixture MDPs, where the transition kernel $P^*$ has a linear function approximation. With this low rank structure, we propose novel adaptive algorithms for model selection, and obtain (order-wise) regret identical to that of an oracle with knowledge of the true model class.

stat.ML↗

arXiv2021

Communication-Efficient and Byzantine-Robust Distributed Learning with Error Feedback

We develop a communication-efficient distributed learning algorithm that is robust against Byzantine worker machines. We propose and analyze a distributed gradient-descent algorithm that performs a simple thresholding based on gradient norms to mitigate Byzantine failures. We show the (statistical) error-rate of our algorithm matches that of Yin et al.~\cite{dong}, which uses more complicated schemes (coordinate-wise median, trimmed mean). Furthermore, for communication efficiency, we consider a generic class of $δ$-approximate compressors from Karimireddi et al.~\cite{errorfeed} that encompasses sign-based compressors and top-$k$ sparsification. Our algorithm uses compressed gradients and gradient norms for aggregation and Byzantine removal respectively. We establish the statistical error rate for non-convex smooth loss functions. We show that, in certain range of the compression factor $δ$, the (order-wise) rate of convergence is not affected by the compression operation. Moreover, we analyze the compressed gradient descent algorithm with error feedback (proposed in \cite{errorfeed}) in a distributed setting and in the presence of Byzantine worker machines. We show that exploiting error feedback improves the statistical error rate. Finally, we experimentally validate our results and show good performance in convergence for convex (least-square regression) and non-convex (neural network training) problems.

cs.LG↗

arXiv2021

An Efficient Framework for Clustered Federated Learning

We address the problem of federated learning (FL) where users are distributed and partitioned into clusters. This setup captures settings where different groups of users have their own objectives (learning tasks) but by aggregating their data with others in the same cluster (same learning task), they can leverage the strength in numbers in order to perform more efficient federated learning. For this new framework of clustered federated learning, we propose the Iterative Federated Clustering Algorithm (IFCA), which alternately estimates the cluster identities of the users and optimizes model parameters for the user clusters via gradient descent. We analyze the convergence rate of this algorithm first in a linear model with squared loss and then for generic strongly convex and smooth loss functions. We show that in both settings, with good initialization, IFCA is guaranteed to converge, and discuss the optimality of the statistical error rate. In particular, for the linear model with two clusters, we can guarantee that our algorithm converges as long as the initialization is slightly better than random. When the clustering structure is ambiguous, we propose to train the models by combining IFCA with the weight sharing technique in multi-task learning. In the experiments, we show that our algorithm can succeed even if we relax the requirements on initialization with random initialization and multiple restarts. We also present experimental results showing that our algorithm is efficient in non-convex problems such as neural networks. We demonstrate the benefits of IFCA over the baselines on several clustered FL benchmarks.

stat.ML↗

arXiv2021

BEAR: Sketching BFGS Algorithm for Ultra-High Dimensional Feature Selection in Sublinear Memory

We consider feature selection for applications in machine learning where the dimensionality of the data is so large that it exceeds the working memory of the (local) computing machine. Unfortunately, current large-scale sketching algorithms show poor memory-accuracy trade-off due to the irreversible collision and accumulation of the stochastic gradient noise in the sketched domain. Here, we develop a second-order ultra-high dimensional feature selection algorithm, called BEAR, which avoids the extra collisions by storing the second-order gradients in the celebrated Broyden-Fletcher-Goldfarb-Shannon (BFGS) algorithm in Count Sketch, a sublinear memory data structure from the streaming literature. Experiments on real-world data sets demonstrate that BEAR requires up to three orders of magnitude less memory space to achieve the same classification accuracy compared to the first-order sketching algorithms. Theoretical analysis proves convergence of BEAR with rate O(1/t) in t iterations of the sketched algorithm. Our algorithm reveals an unexplored advantage of second-order optimization for memory-constrained sketching of models trained on ultra-high dimensional data sets.

cs.LG↗

arXiv2021

Training Recommender Systems at Scale: Communication-Efficient Model and Data Parallelism

In this paper, we consider hybrid parallelism -- a paradigm that employs both Data Parallelism (DP) and Model Parallelism (MP) -- to scale distributed training of large recommendation models. We propose a compression framework called Dynamic Communication Thresholding (DCT) for communication-efficient hybrid training. DCT filters the entities to be communicated across the network through a simple hard-thresholding function, allowing only the most relevant information to pass through. For communication efficient DP, DCT compresses the parameter gradients sent to the parameter server during model synchronization. The threshold is updated only once every few thousand iterations to reduce the computational overhead of compression. For communication efficient MP, DCT incorporates a novel technique to compress the activations and gradients sent across the network during the forward and backward propagation, respectively. This is done by identifying and updating only the most relevant neurons of the neural network for each training sample in the data. We evaluate DCT on publicly available natural language processing and recommender models and datasets, as well as recommendation systems used in production at Facebook. DCT reduces communication by at least $100\times$ and $20\times$ during DP and MP, respectively. The algorithm has been deployed in production, and it improves end-to-end training time for a state-of-the-art industrial recommender model by 37\%, without any loss in performance.

cs.LG↗

arXiv2021

LocalNewton: Reducing Communication Bottleneck for Distributed Learning

To address the communication bottleneck problem in distributed optimization within a master-worker framework, we propose LocalNewton, a distributed second-order algorithm with local averaging. In LocalNewton, the worker machines update their model in every iteration by finding a suitable second-order descent direction using only the data and model stored in their own local memory. We let the workers run multiple such iterations locally and communicate the models to the master node only once every few (say L) iterations. LocalNewton is highly practical since it requires only one hyperparameter, the number L of local iterations. We use novel matrix concentration-based techniques to obtain theoretical guarantees for LocalNewton, and we validate them with detailed empirical evaluation. To enhance practicability, we devise an adaptive scheme to choose L, and we show that this reduces the number of local iterations in worker machines between two model synchronizations as the training proceeds, successively refining the model quality at the master. Via extensive experiments using several real-world datasets with AWS Lambda workers and an AWS EC2 master, we show that LocalNewton requires fewer than 60% of the communication rounds (between master and workers) and less than 40% of the end-to-end running time, compared to state-of-the-art algorithms, to reach the same training~loss.

cs.DC↗

arXiv2021

Byzantine-Robust Distributed Learning: Towards Optimal Statistical Rates

In large-scale distributed learning, security issues have become increasingly important. Particularly in a decentralized environment, some computing units may behave abnormally, or even exhibit Byzantine failures -- arbitrary and potentially adversarial behavior. In this paper, we develop distributed learning algorithms that are provably robust against such failures, with a focus on achieving optimal statistical performance. A main result of this work is a sharp analysis of two robust distributed gradient descent algorithms based on median and trimmed mean operations, respectively. We prove statistical error rates for three kinds of population loss functions: strongly convex, non-strongly convex, and smooth non-convex. In particular, these algorithms are shown to achieve order-optimal statistical error rates for strongly convex losses. To achieve better communication efficiency, we further propose a median-based distributed algorithm that is provably robust, and uses only one communication round. For strongly convex quadratic loss, we show that this algorithm achieves the same optimal error rate as the robust distributed gradient descent algorithms.

cs.LG↗

arXiv2021

Provably Breaking the Quadratic Error Compounding Barrier in Imitation Learning, Optimally

We study the statistical limits of Imitation Learning (IL) in episodic Markov Decision Processes (MDPs) with a state space $\mathcal{S}$. We focus on the known-transition setting where the learner is provided a dataset of $N$ length-$H$ trajectories from a deterministic expert policy and knows the MDP transition. We establish an upper bound $O(|\mathcal{S}|H^{3/2}/N)$ for the suboptimality using the Mimic-MD algorithm in Rajaraman et al (2020) which we prove to be computationally efficient. In contrast, we show the minimax suboptimality grows as $Ω( H^{3/2}/N)$ when $|\mathcal{S}|\geq 3$ while the unknown-transition setting suffers from a larger sharp rate $Θ(|\mathcal{S}|H^2/N)$ (Rajaraman et al (2020)). The lower bound is established by proving a two-way reduction between IL and the value estimation problem of the unknown expert policy under any given reward function, as well as building connections with linear functional estimation with subsampled observations. We further show that under the additional assumption that the expert is optimal for the true reward function, there exists an efficient algorithm, which we term as Mimic-Mixture, that provably achieves suboptimality $O(1/N)$ for arbitrary 3-state MDPs with rewards only at the terminal layer. In contrast, no algorithm can achieve suboptimality $O(\sqrt{H}/N)$ with high probability if the expert is not constrained to be optimal. Our work formally establishes the benefit of the expert optimal assumption in the known transition setting, while Rajaraman et al (2020) showed it does not help when transitions are unknown.

cs.LG↗

arXiv2021

Boundary thickness and robustness in learning models

Robustness of machine learning models to various adversarial and non-adversarial corruptions continues to be of interest. In this paper, we introduce the notion of the boundary thickness of a classifier, and we describe its connection with and usefulness for model robustness. Thick decision boundaries lead to improved performance, while thin decision boundaries lead to overfitting (e.g., measured by the robust generalization gap between training and testing) and lower robustness. We show that a thicker boundary helps improve robustness against adversarial examples (e.g., improving the robust test accuracy of adversarial training) as well as so-called out-of-distribution (OOD) transforms, and we show that many commonly-used regularization and data augmentation procedures can increase boundary thickness. On the theoretical side, we establish that maximizing boundary thickness during training is akin to the so-called mixup training. Using these observations, we show that noise-augmentation on mixup training further increases boundary thickness, thereby combating vulnerability to various forms of adversarial attacks and OOD transforms. We can also show that the performance improvement in several lines of recent work happens in conjunction with a thicker boundary.

cs.LG↗

arXiv2020

CoVer: Collaborative Light-Node-Only Verification and Data Availability for Blockchains

Validating a blockchain incurs heavy computation, communication, and storage costs. As a result, clients with limited resources, called light nodes, cannot verify transactions independently and must trust full nodes, making them vulnerable to security attacks. Motivated by this problem, we ask a fundamental question: can light nodes securely validate without any full nodes? We answer affirmatively by proposing CoVer, a decentralized protocol that allows a group of light nodes to collaboratively verify blocks even under a dishonest majority, achieving the same level of security for block validation as full nodes while only requiring a fraction of the work. In particular, work per node scales down proportionally with the number of participants (up to a log factor), resulting in computation, communication, and storage requirements that are sublinear in block size. Our main contributions are light-node-only protocols for fraud proofs and data availability.

cs.CR↗

arXiv2020

FastSecAgg: Scalable Secure Aggregation for Privacy-Preserving Federated Learning

Recent attacks on federated learning demonstrate that keeping the training data on clients' devices does not provide sufficient privacy, as the model parameters shared by clients can leak information about their training data. A 'secure aggregation' protocol enables the server to aggregate clients' models in a privacy-preserving manner. However, existing secure aggregation protocols incur high computation/communication costs, especially when the number of model parameters is larger than the number of clients participating in an iteration -- a typical scenario in federated learning. In this paper, we propose a secure aggregation protocol, FastSecAgg, that is efficient in terms of computation and communication, and robust to client dropouts. The main building block of FastSecAgg is a novel multi-secret sharing scheme, FastShare, based on the Fast Fourier Transform (FFT), which may be of independent interest. FastShare is information-theoretically secure, and achieves a trade-off between the number of secrets, privacy threshold, and dropout tolerance. Riding on the capabilities of FastShare, we prove that FastSecAgg is (i) secure against the server colluding with 'any' subset of some constant fraction (e.g. $\sim10\%$) of the clients in the honest-but-curious setting; and (ii) tolerates dropouts of a 'random' subset of some constant fraction (e.g. $\sim10\%$) of the clients. FastSecAgg achieves significantly smaller computation cost than existing schemes while achieving the same (orderwise) communication cost. In addition, it guarantees security against adaptive adversaries, which can perform client corruptions dynamically during the execution of the protocol.

cs.CR↗

arXiv2020

OverSketched Newton: Fast Convex Optimization for Serverless Systems

Motivated by recent developments in serverless systems for large-scale computation as well as improvements in scalable randomized matrix algorithms, we develop OverSketched Newton, a randomized Hessian-based optimization algorithm to solve large-scale convex optimization problems in serverless systems. OverSketched Newton leverages matrix sketching ideas from Randomized Numerical Linear Algebra to compute the Hessian approximately. These sketching methods lead to inbuilt resiliency against stragglers that are a characteristic of serverless architectures. Depending on whether the problem is strongly convex or not, we propose different iteration updates using the approximate Hessian. For both cases, we establish convergence guarantees for OverSketched Newton and empirically validate our results by solving large-scale supervised learning problems on real-world datasets. Experiments demonstrate a reduction of ~50% in total running time on AWS Lambda, compared to state-of-the-art distributed optimization schemes.

cs.DC↗

arXiv2020

Alternating Minimization Converges Super-Linearly for Mixed Linear Regression

We address the problem of solving mixed random linear equations. We have unlabeled observations coming from multiple linear regressions, and each observation corresponds to exactly one of the regression models. The goal is to learn the linear regressors from the observations. Classically, Alternating Minimization (AM) (which is a variant of Expectation Maximization (EM)) is used to solve this problem. AM iteratively alternates between the estimation of labels and solving the regression problems with the estimated labels. Empirically, it is observed that, for a large variety of non-convex problems including mixed linear regression, AM converges at a much faster rate compared to gradient based algorithms. However, the existing theory suggests similar rate of convergence for AM and gradient based methods, failing to capture this empirical behavior. In this paper, we close this gap between theory and practice for the special case of a mixture of $2$ linear regressions. We show that, provided initialized properly, AM enjoys a \emph{super-linear} rate of convergence in certain parameter regimes. To the best of our knowledge, this is the first work that theoretically establishes such rate for AM. Hence, if we want to recover the unknown regressors upto an error (in $\ell_2$ norm) of $ε$, AM only takes $\mathcal{O}(\log \log (1/ε))$ iterations. Furthermore, we compare AM with a gradient based heuristic algorithm empirically and show that AM dominates in iteration complexity as well as wall-clock time.

stat.ML↗