Abstract

Sparse neural networks are shown to give accurate predictions competitive to denser versions, while also minimizing the number of arithmetic operations performed. However current GPU hardware can only exploit structured sparsity patterns for better efficiency. We propose a framework for generating structured multilevel block sparse neural networks by using the theory of graph products. Our Ramanujan bipartite graph product (RBGP) framework uses products of Ramanujan graphs to obtain the best connectivity for a given level of sparsity. This essentially ensures that the i.) the networks has the structured block sparsity for which runtime efficient algorithms exists, ii.) the model gives high prediction accuracy, due to the better expressive power derived from the connectivity of the graph, iii.) the graph data structure has a succinct representation that can be stored efficiently in memory. We use our framework to design a specific connectivity pattern called RBGP4 which makes efficient use of the memory hierarchy available on GPU. We benchmark our approach on image classification and machine translation tasks with an edge (Jetson Nano 2GB) as well as server (V100) GPUs. When compared with commonly used sparsity patterns like unstructured and block, we obtain significant speedups while achieving the same level of accuracy.

Abstract

Sparse neural networks have been proven to generate efficient and better runtimes when compared to dense neural networks. Accelera- tion in runtime is better achieved with structured sparsity. However, generating an efficient sparsity structure to maintain both runtime and accuracy is a challenging task. In this paper, we implement the RBGP4 sparsity pattern derived from the Ramanujan Bipartite Graph Product (RBGP) framework on various Computer Vision tasks and test how well it performs w.r.t accuracy and runtime. Us- ing this approach, we generate structured sparse neural networks which has multiple levels of block sparsity that generates good connectivity due to the presence of Ramanujan bipartite graphs. We benchmark our approach on Semantic Segmentation and Pose Estimation tasks on an edge device (Jetson Nano 2GB) as well as server (V100) GPUs. We compare the results obtained for RBGP4 sparsity pattern with the unstructured and block sparsity patterns. When compared to sparsity patterns like unstructured and block, we obtained significant speedups while maintaining accuracy. KEYWORDS Sparse Neural Networks, Ramanujan Graphs, Semantic Segmenta- tion, Pose Detection ACM Reference Format: Thejasvi Konduru, Furqan Ahmed Shaik, Girish Varma, and Kishore Kotha- palli. 2023. Accelerating Computer Vision Tasks on GPUs using Ramanujan Graph Product Framework. In 6th Joint International Conference on Data Science & Management of Data (10th ACM IKDD CODS and 28th COMAD) (CODS-COMAD 2023), January 4–7, 2023, Mumbai, India. ACM, New York, NY, USA, 5 pages. https://doi.org/10.1145/3570991.3571044

Abstract

The design and implementation of parallel algorithms for dynamic graph problems is attracting significant research attention in the recent years, driven by numerous applications to social network analysis, neuroscience, and protein interaction networks. One such problem is the computation of PageRank values of vertices in a directed graph. This paper presents two new parallel algorithms for recomputing the PageRank values of vertices in a dynamic graph. Our techniques require the recomputation of the PageRank of only the vertices affected by the insertion/deletion of a batch of edges. We conduct detailed experimental studies of our algorithm on a set of 11 realworld graphs. Our results on Intel Xeon Silver 4116 CPU and NVIDIA Tesla V100 PCIe 16GB GPU indicate that our algorithms outperform static and dynamic update algorithms by 6.1×and 8.6×on the CPU, and by 9.8×and 9.3×on the GPU respectively. We also compare the performance of the algorithms in batched mode to cumulative single-edge updates.

Abstract

Finding the biconnected components of a graph has a large number of applications in many other graph problems including planarity testing, computing the centrality metrics, finding the (weighted) vertex cover, coloring, and the like. Recent years saw the design of efficient algorithms for this problem across sequential and parallel computational models. However, current algorithms do not work in the setting where the underlying graph changes over time in a dynamic manner via the insertion or deletion of edges. Dynamic algorithms in the sequential setting that obtain the biconnected components of a graph upon insertion or deletion of a single edge are known from over two decades ago. Parallel algorithms for this problem are not heavily studied. In this paper, we design shared-memory parallel algorithms that obtain the biconnected components of a graph subsequent to the insertion or deletion of a batch of edges

Abstract

2022 IEEE International Conference on Blockchain (Blockchain) BLOC KVAC: A Un iversally Acceptable and Ideal Vaccination System on Blockchain Manik a Sharma Center for Securit y, Theory, and Algorithmic Research Intern ationa l Institute of Inform ation Technology, Hyderabad Gach ibowli, Hyderabad, India 500 032 manik a.sharm a @research.iiit.ac.in Kishore Kothap alli Cent er for Securit y, Theo ry, and Algorithmic Research Interna tional Institute of Information Technology, Hyderabad Gachibowli, Hyderabad , Indi a 500 032 kisho re @iiit.ac .in Sujit Gujar Machin e Learning Lab International Institute of Information Technology, Hyderabad Gachib owli, Hyderabad , India 500 032 suj it.gujar @iiit.ac.in Abstract-Vaccines are essential in pro tecting indivi dua ls and commun ities from the risk of ha rmful an d fatal diseases. The wor ld is migra ting towards the digitization of vacci nation docu- mentation. Still, the current digita l vacci nation is a centra lized mode l, lacking a nnive rsal acceptance. A Universally Acceptable an d Id eal Vacci nat ion System (UAIVS ) sho nld possess (a) a sta ndard vaccination cert ificate ver ificat ion algorithm, (h) p r i- vacy to per sonal in form a tion, (c) sca lab ilit y, (d) s npport lor all vacci nat ions, (e) acco u ntab ility for vaccination information access, and (f) a sco pe for scientific researc h a nd d evelopm en t. Blockc ha in tec hno logy int ro dnces d ecentr alizat ion, immn tability, acco unta bility, per sistence, and liveliness. T hese pro per ties ma ke block cha in an ideal can di da te to solve the existing pro blems in digital vacci nat ion. In this pa per, we prop ose a novel blockchain- base d syste m, n am ely BLO CKVAC, to implem ent a UA IVS . F urt her more , o ur wor k provi des a way to track an individ nal's an d fa mily's vacc ination histo ry, which is esse ntial for sc ient ific research a nd developm ent . BLOCKVAC is highly modular, sca l- a ble, an d cost -efficien t, pr oviding eno ugh roo m for fnt n re feature addi tio ns. The c urre nt works lack cost and sca lab ility ana lysis. WI' show thc scalability of o ur system t hro ugh a th oron gh a nalys is. O ur impl em entation shows t hat add ing one vacc inati on re cor d costs 7 . l(j ~2 ~ x 10- 6 ETH ( ~ U.U26US D) for reg istratio n, 2.5089 x 10- 5 ETH ( ~ 0.07 8USD) for vaccina tion, a nd th e syste m ca n easi ly han dl e np to 20K vaccinations per hon r. Ind ex Terms- B1ockc ha in, Eth ere um, Sm art Contrac ts, Public Acco unta bility, Sca lab ilit y I. I N TR O D UC TI O N The vaccination process, also known as imm unization , was introduced in the late 18th century by the British physician Edward Jenn er against the deadly smallpox virus [ I]. Since then. it has become an inevitab le medical proc edur e, protecting individuals and communities from harmful and fatal diseases. The World Hea lth Organi zation (WHO) estimates that vac- cination saves around 2.5 millio n child lives every year [2]. A record of vacci nation details massively aid s in providing qualit y health care to individuals [3]. Whi le secure storage and verificat ion of vaccination details are challenges even within a country, it is tediou s to merge them from across all the countries [4 ]. Moreover. a definitive process is the need of the hour to verify the required vaccinati on certification for travelers when they enter a country. We wou ld like to than k MeitY and Ripp le for fundin g this research . O ver view of existing vac cina tion mod els: Traditional paper- based vaccination records [5] are prone to challenges such as the possibil ity of losing, damaging, or forging the records without any accountability. Digital solutions replace the need for paper, provide an immutable card, and reduc e manual labor [6]. However, the current vacci nation systems follow a centralized model, where the users need to trust the cent ral autho rity. Further. these sys tems are prone to single-point failures and jeopardize data security P 1. Moreover, current vaccination portal s mainly deal with just slot bookings and issuin g non -verifiab le vaccination certificates. Mo tiva tion: Th e Covid-19 pandemic has caused a digital revolution in vaccination data managem ent. Countri es are shifting towards a digital solution to secure vaccination data ami issue certificates. However, such a system should be universally acceptabl e. We identi fy the properties requ ired for such a sys tem and call it a Universally Acceptable and Ideal Vaccination System (UA IVS). We furt her categorize the proper- ties of UAIVS into two subcategories: Universally Acceptable Vaccination System and Ideal Va ccination System. A Universally Acce ptable Vaccination System co

Abstract

Welcome to virtual HiPC 2021, the 28th edition of the IEEE International Conference on High Performance Computing, Data, and Analytics. It serves as a forum to present current work by accomplished HPC researchers from around the globe, to explore developing topics in the field, and to highlight related activities in Asia.

Abstract

— Centrality measures on graphs have found applications in a large number of domains including modeling the spread of an infection/disease, social network analysis, and transportation networks. As a result, parallel algorithms for computing various centrality metrics on graphs are gaining significant research attention in recent years. In this paper, we study parallel algorithms for the percolation centrality measure which extends the betweenness-centrality measure by incorporating a time dependent state variable with every node. We present parallel algorithms that compute the source-based and source-destination variants of the percolation centrality values of nodes in a network. Our algorithms extend the algorithm of Brandes, introduce optimizations aimed at exploiting the structural properties of graphs, and extend the algorithmic techniques introduced by Sariyuce et al. [26] in the context of centrality computation. Experimental studies of our algorithms on an Intel Xeon(R) Silver 4116 CPU and an Nvidia Tesla V100 GPU on a collection of 12 real-world graphs indicate that our algorithmic techniques offer a significant speedup.

Abstract

Centrality measures on graphs have found appli- cations in a large number of domains including modeling the spread of an infection/disease, social network analysis, and transportation networks. As a result, parallel algorithms for computing various centrality metrics on graphs are gaining significant research attention in recent years. In this paper, we study parallel algorithms for the percolation centrality measure which extends the betweenness-centrality measure by incorporating a time dependent state variable with every node. We present parallel algorithms that compute the source-based and source-destination variants of the percolation centrality values of nodes in a network. Our algorithms extend the algorithm of Brandes, introduce optimizations aimed at exploiting the structural properties of graphs, and extend the algorithmic techniques introduced by Sariyuce et al. [26] in the context of centrality computation. Experimental studies of our algorithms on an Intel Xeon(R) Silver 4116 CPU and an Nvidia Tesla V100 GPU on a collection of 12 real-world graphs indicate that our algorithmic techniques offer a significant speedup.

Abstract

— The effect of adjusting damping factor α and tolerance τ on iterations needed for PageRank computation is studied here. Relative performance of PageRank computation with L1, L2, and L∞ norms used as convergence check, are also compared with six possible mean ratios. It is observed that increasing the damping factor α linearly increases the iterations needed almost exponentially. On the other hand, decreasing the tolerance τ exponentially decreases the iterations needed almost exponentially. On average, PageRank with L∞ norm as convergence check is the fastest, quickly followed by L2 norm, and then L1 norm. For large graphs, above certain tolerance τ values, convergence can occur in a single iteration. On the contrary, below certain tolerance τ values, sensitivity issues can begin to appear, causing computation to halt at maximum iteration limit without convergence. The six mean ratios for relative performance comparison are based on arithmetic, geometric, and harmonic mean, as well as the order of ratio calculation. Among them GM-RATIO, geometric mean followed by ratio calculation, is found to be most stable, followed by AM-RATIO.

Abstract

The effect of adjusting damping factor α, from a small initial value α0 to the final desired αf value, upon then iterations needed for PageRank computation is observed. Adjustment of the damping factor is done in one or more steps. Results show no improvement in performance over a fixed damping factor based PageRank.