1<!DOCTYPE html> 2<html> 3 4 <head> 5 <meta charset="utf-8"> 6 <meta http-equiv="X-UA-Compatible" content="IE=edge"> 7 <meta name="viewport" content="width=device-width, initial-scale=1"> 8 <title>Intel Multimodal AI Innovation Lab - Publications</title> 9 <meta name="description" content="Intel Multimodal AI Innovation Lab -- Publications."> 10 <link rel="stylesheet" href="/css/main.css"> 11 <link rel="canonical" href="/publications/"> 12<link rel="shortcut icon" type ="image/x-icon" href="/images/favicon.ico"> 13 14
15<script>
vendor: 345 bytes, lines 15-21
15 16 (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){ 17 (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o), 18 m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m) 19 })(window,document,'script','https://www.google-analytics.com/analytics.js','ga'); 20 21 ga('create', '
21UA-82472331-1
vendor: 39 bytes, lines 21-23
21', 'auto'); 22 ga('send', 'pageview'); 23
24</script>
24 25 26 27 28</head> 29 30 31 <body> 32 33 <div class="navbar navbar-default navbar-fixed-top" role="navigation"> 34 <div class="container-fluid"> 35 <div class="navbar-header"> 36 <button type="button" class="navbar-toggle collapsed" data-toggle="collapse" data-target="#navbar-collapse-1" aria-expanded="false"> 37 <span class="sr-only">Toggle navigation</span> 38 <span class="icon-bar"></span> 39 <span class="icon-bar"></span> 40 <span class="icon-bar"></span> 41 </button> 42 43 <a class="navbar-brand" href="/">Intel Multimodal AI Innovation Lab</a> 44 <!-- <img src="/images/logopic/ailab.png" style="width: 80px"> --> 45 </div> 46 <div class="collapse navbar-collapse" id="navbar-collapse-1"> 47 <ul class="nav navbar-nav navbar-right"> 48 <li><a href="/">Home</a></li> 49 <li><a href="/team">Team</a></li> 50 <li><a href="/allblogs">Blogs</a></li> 51 <li><a href="/publications">Publications</a></li> 52 <li><a href="/research">Research</a></li> 53 <li><a href="/pictures">(Pics)</a></li> 54 </ul> 55 </div> 56 </div> 57</div> 58 59 60 <div class="container-fluid"> 61 <div class="row"> 62 <div id="gridid" class="col-sm-12"> 63 <h1 id="-publications">ð Publications</h1> 64 65<!-- ## Group highlights 66 67**At the end of this page, you can find the [full list of publications and patents](#full-list-of-publications). All papers are also available on [arXiv](https://arxiv.org/search/?searchtype=author&query=Allan%2C+M+P).** 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387<p> </p> 388 389 390## Patents 391<em>Milan P Allan, S Gröblacher, RA Norte, M Leeuwenhoek</em><br />Novel atomic force microscopy probes with phononic crystals<br /> PCT/NL20-20/050797 (2020) 392 393<em>Milan P Allan</em><br /> Methods of manufacturing superconductor and phononic elements <br /> <a href="https://patents.google.com/patent/US10439125B2/en?inventor=Milan+ALLAN&oq=inventor:(Milan+ALLAN)">US10439125B2 (2016)</a> --> 394 395<!-- ## Publications --> 396 397<!-- 398 399 Advancing Reference-free Evaluation of Video Captions with Factual Analysis <br /> 400 <em> </em><br /><a href="">ACL</a> 401 402 403 404 Harnessing Object Grounding for Time-Sensitive Video Understanding <br /> 405 <em> </em><br /><a href="">WACV</a> 406 407 408 409 EASG-Bench: Video Q&A Benchmark with Egocentric Action Scene Graphs <br /> 410 <em> </em><br /><a href="">ICCVW</a> 411 412 413 414 ByDeWay: Boost Your multimodal LLM with DEpth prompting in a training-free Way <br /> 415 <em> </em><br /><a href="">ICCVW</a> 416 417 418 419 Toward Scalable Video Narration: A Training-free Approach Using Multimodal Large Language Models <br /> 420 <em> </em><br /><a href="">ICCVW</a> 421 422 423 424 MoDM: Efficient Serving for Image Generation via Mixture-of-Diffusion Models <br /> 425 <em> </em><br /><a href="">ASPLOS</a> 426 427 428 429 ESSENTIAL: Episodic and Semantic Memory Integration for Video Class-Incremental Learning <br /> 430 <em> </em><br /><a href="">ICCV</a> 431 432 433 434 OuroMamba: A Data-Free Quantization Framework for Vision Mamba Models <br /> 435 <em> </em><br /><a href="">ICCV</a> 436 437 438 439 CITER: Collaborative Inference for Efficient Large Language Model Decoding with Token-Level Routing <br /> 440 <em> </em><br /><a href="">COLM</a> 441 442 443 444 SEAL: Steerable reasoning calibration of large language models for free <br /> 445 <em> </em><br /><a href="">COLM</a> 446 447 448 449 PISA Experiments: Exploring Physics Post-Training for Video Diffusion Models by Watching Stuff Drop <br /> 450 <em> </em><br /><a href="">ICML</a> 451 452 453 454 On-the-Fly Adaptive Distillation of Transformer to Dual-State Linear Attention for Long-Context LLM Serving <br /> 455 <em> </em><br /><a href="">ICML</a> 456 457 458 459 PALADIN : Robust Neural Fingerprinting for Text-to-Image Diffusion Models <br /> 460 <em> </em><br /><a href="">CVPRW</a> 461 462 463 464 LAMB: A Training-Free Method to Enhance the Long-Context Understanding of SSMs via Attention-Guided Token Filtering <br /> 465 <em> </em><br /><a href="">ACL</a> 466 467 468 469 Accelerating LLM Inference with Flexible N: M Sparsity via A Fully Digital Compute-in-Memory Accelerator <br /> 470 <em> </em><br /><a href="">DAC</a> 471 472 473 474 LVLM-Compress-Bench: Benchmarking the Broader Impact of Large Vision-Language Model Compression <br /> 475 <em> </em><br /><a href="">NAACL</a> 476 477 478 479 Stiefel Flow Matching for Moment-Constrained Structure Elucidation <br /> 480 <em> </em><br /><a href="">ICLR</a> 481 482 483 484 SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Models <br /> 485 <em> </em><br /><a href="">ICLR</a> 486 487 488 489 MatExpert: Decomposing Materials Discovery By Mimicking Human Experts <br /> 490 <em> </em><br /><a href="">ICLR</a> 491 492 493 494 Scaling Long Context Training Data by Long-Distance Referrals <br /> 495 <em> </em><br /><a href="">ICLR</a> 496 497 498 499 LANTERN: Accelerating Visual Autoregressive Models with Relaxed Speculative Decoding <br /> 500 <em> </em><br /><a href="">ICLR</a> 501 502 503 504 GraphAny: A Foundation Model for Node Classification on Any Graph <br /> 505 <em> </em><br /><a href="">ICLR</a> 506 507 508 509 MambaExtend: A Training-Free Approach to Improve Long Context Extension of Mamba <br /> 510 <em> </em><br /><a href="">ICLR</a> 511 512 513 514 QMambaExtend: Improving Long-Context Extension of Memory-Efficient Mamba Models <br /> 515 <em> </em><br /><a href="">ICLR workshop</a> 516 517 518 519 LANTERN++: Enhanced Relaxed Speculative Decoding with Static Tree Drafting for Visual Auto-regressive Models <br /> 520 <em> </em><br /><a href="">ICLR workshop</a> 521 522 523 524 MaskVD: Region Masking for Efficient Video Object Detection <br /> 525 <em> </em><br /><a href="">WACV</a> 526 527 528 529 Ego-VPA: Egocentric Video Understanding with Parameter-efficient Adaptation <br /> 530 <em> </em><br /><a href="">WACV</a> 531 532 533 534 Deep Geometric Moments Promote Shape Consistency in Text-to-3D Generation <br /> 535 <em> </em><br /><a href="">WACV</a> 536 537 538 539 BASED: Bundle-Adjusting Surgical Endoscopic Dynamic Video Reconstruction using Neural Radiance Fields <br /> 540 <em> </em><br /><a href="">WACV</a> 541 542 543 544 LLaMA-NAS: Efficient Neural Architecture Search for Large Language Models <br /> 545 <em> </em><br /><a href="https://arxiv.org/abs/2405.18377">ECCV</a> 546 547 548 549 Bit-by-Bit: Investigating the Vulnerabilities of Binary Neural Networks to Adversarial Bit Flipping <br /> 550 <em> </em><br /><a href="">TMLR</a> 551 552 553 554 R.A.C.E.: Robust Adversarial Concept Erasure for Secure Text-to-Image Diffusion Model <br /> 555 <em> </em><br /><a href="">ECCV</a> 556 557 558 559 CLAMP-ViT: Contrastive Data-Free Learning for Adaptive Post-Training Quantization of ViTs <br /> 560 <em> </em><br /><a href="">ECCV</a> 561 562 563 564 GenQ: Quantization in Low Data Regimes with Generative Synthetic Data <br /> 565 <em> </em><br /><a href="https://arxiv.org/pdf/2312.05272">ECCV</a> 566 567 568 569 Sensi-BERT: Towards Sensitivity Driven Fine-Tuning for Parameter Efficient BERT <br /> 570 <em> </em><br /><a href="https://arxiv.org/abs/2307.11764">ICASSP</a> 571 572 573 574 SimQ-NAS: Simultaneous Quantization Policy and Neural Architecture Search <br /> 575 <em> </em><br /><a href="https://arxiv.org/abs/2312.13301">arXiv</a> 576 577 578 579 Searching the Space of High-Value Molecules Using Reinforcement Learning and Language Models <br /> 580 <em> </em><br /><a href="">ICLR</a> 581 582 583 584 Towards Foundation Models for Knowledge Graph Reasoning <br /> 585 <em> </em><br /><a href="">ICLR</a> 586 587 588 589 Fusing Models with Complementary Expertise <br /> 590 <em> </em><br /><a href="https://openreview.net/pdf?id=PhMrGCMIRL">ICLR</a> 591 592 593 594 AFLoRA: Adaptive Freezing of Low-Rank Adapters for Parameter Efficient Fine-Tuning of Language Models <br /> 595 <em> </em><br /><a href="https://arxiv.org/pdf/2403.13269.pdf">ACL</a> 596 597 598 599 RLNet: Robust Linearized Networks for Efficient Private Inference <br /> 600 <em> </em><br /><a href="">CVPRW</a> 601 602 603 604 Block Selective Reprogramming for On-device Training of Vision Transformers <br /> 605 <em> </em><br /><a href="">CVPRW</a> 606 607 608 609 VideoSAGE: Video Summarization with Graph Representation Learning <br /> 610 <em> </em><br /><a href="https://arxiv.org/pdf/2404.10539.pdf">CVPRW</a> 611 612 613 614 LAVITI: Contrastive Language Video Time Pre-training <br /> 615 <em> </em><br /><a href="">CVPRW</a> 616 617 618 619 Pruning Small Magnitude Weights Monotonically and Irreversibly Impairs Difficult Downstream Tasks in LLMs <br /> 620 <em> </em><br /><a href="">arXiv</a> 621 622 623 624 Towards Joint Optimization for DNN Architecture and Configuration for Compute-In-Memory Hardware <br /> 625 <em> </em><br /><a href="https://arxiv.org/abs/2402.11780">arXiv</a> 626 627 628 629 Ego-VPA: Egocentric Video Understanding with Parameter-efficient Adaptation <br /> 630 <em> </em><br /><a href="">Under review</a> 631 632 633 634 PASTE: Prompt-Aligned Scene TExture synthesis <br /> 635 <em> </em><br /><a href="">Under review</a> 636 637 638 639 Image Sculpting: Precise Object Editing with 3D Geometry Control <br /> 640 <em> </em><br /><a href="https://arxiv.org/abs/2401.01702">CVPR</a> 641 642 643 644 WOUAF: Weight Modulation for User Attribution and Fingerprinting in Text-to-Image Diffusion Models <br /> 645 <em> </em><br /><a href="https://arxiv.org/abs/2306.04744">CVPR</a> 646 647 648 649 Action Scene Graphs for Long-Form Understanding of Egocentric Videos <br /> 650 <em> </em><br /><a href="https://arxiv.org/abs/2312.03391">CVPR</a> 651 652 653 654 BASED: Bundle-Adjusting Surgical Endoscopic Dynamic Video Reconstruction using Neural Radiance Fields <br /> 655 <em> </em><br /><a href="https://arxiv.org/abs/2309.15329">ArXiv</a> 656 657 658 659 Joint Hand Motion and Interaction Hotspots Prediction from Egocentric Videos <br /> 660 <em> </em><br /><a href="https://arxiv.org/abs/2204.01696">arXiv</a> 661 662 663 664 Text Spotting Transformers <br /> 665 <em> </em><br /><a href="https://arxiv.org/abs/2204.01918">arXiv</a> 666 667 668 669 Learning long-term spatial-temporal graphs for active speaker detection <br /> 670 <em> </em><br /><a href="https://arxiv.org/pdf/2207.07783.pdf">arXiv</a> 671 672 673 674 Single-Stage Visual Relationship Learning using Conditional Queries <br /> 675 <em> </em><br /><a href="https://openreview.net/forum?id=hPVXHzzK0z">openReview</a> 676 677 678 679 STHG: Spatial-Temporal Heterogeneous Graph Learning for Advanced Audio-Visual Diarization <br /> 680 <em> </em><br /><a href="https://arxiv.org/abs/2306.10608">ArXiv</a> 681 682 683
684 BAA-NGP: Bundle-Adjusting Accelerated Neural Graphics Primitives <br /> 685 <em> </em><br /><a href="https://arxiv.org/abs/2306.04166">ArXiv</a> 686 687 688 689 SViTT: Temporal Learning of Sparse Video-Text Transformer <br /> 690 <em> </em><br /><a href="https://arxiv.org/abs/2304.08809">arXiv</a> 691 692 693 694 Unbiased video scene graph generation <br /> 695 <em> </em><br /><a href="https://arxiv.org/abs/2304.00733">arXiv</a> 696 697 698 699 Exploiting Long-Term Dependencies for Generating Dynamic Scene Graphs <br /> 700 <em> </em><br /><a href="https://openaccess.thecvf.com/content/WACV2023/papers/Feng_Exploiting_Long-Term_Dependencies_for_Generating_Dynamic_Scene_Graphs_WACV_2023_paper.pdf">CVF</a> 701 702 703 704 TrimBERT: Tailoring BERT for Trade-offs <br /> 705 <em> </em><br /><a href="https://arxiv.org/abs/2202.12411">arXiv</a> 706 707 708 709 Accelerating Neural Architecture Exploration Across Modalities Using Genetic Algorithms <br /> 710 <em> </em><br /><a href="https://arxiv.org/abs/2202.12934">arXiv</a> 711 712 713 714 A Hardware-Aware Framework for Accelerating Neural Architecture Search Across Modalities <br /> 715 <em> </em><br /><a href="https://arxiv.org/abs/2205.10358">arXiv</a> 716 717 718 719 Automated Super-Network Generation for Scalable Neural Architecture Search <br /> 720 <em> </em><br /><a href="https://openreview.net/forum?id=HK-zmbTB8gq">OpenReview</a> 721 722 723 724 FLOAT: Fast Learnable Once-for-All Adversarial Training for Tunable Trade-off between Accuracy and Robustness <br /> 725 <em> </em><br /><a href="https://openreview.net/forum?id=MXrIVw-F_a4">OpenReview</a> 726 727 728 729 Sparse Mixture Once-for-all Adversarial Training for Efficient in-situ Trade-off between Accuracy and Robustness of DNNs <br /> 730 <em> </em><br /><a href="https://arxiv.org/abs/2302.03523">arXiv</a> 731 732 733 734 AttentionLite: Towards Efficient Self-attention models for Vision <br /> 735 <em> </em><br /><a href="https://arxiv.org/abs/2101.05216">arXiv</a> 736 737 738 739 Improving Accuracy of Respiratory Rate Estimation by Restoring High Resolution Features With Transformers and Recursive Convolutional Models <br /> 740 <em> </em><br /><a href="https://openaccess.thecvf.com/content/CVPR2021W/CVPM/html/Kwasniewska_Improving_Accuracy_of_Respiratory_Rate_Estimation_by_Restoring_High_Resolution_CVPRW_2021_paper.html">CVF</a> 741 742 743 744 Undivided Attention: Are Intermediate Layers Necessary for BERT? <br /> 745 <em> </em><br /><a href="https://neurips2021-nlp.github.io/papers/33/CameraReady/Undivided_Attention_Neurips_Workshop_2021_CameraReady.pdf">NeurIPS</a> 746 747 748 749 Deep-SWIM: A few-shot learning approach to classify Solar WInd Magnetic field structures <br /> 750 <em> </em><br /><a href="https://ml4physicalsciences.github.io/2021/files/NeurIPS_ML4PS_2021_138.pdf">NeurIPS</a> 751 752 753 754 Self-Attentive Pooling for Efficient Deep Learning <br /> 755 <em> </em><br /><a href="https://openaccess.thecvf.com/content/WACV2023/papers/Chen_Self-Attentive_Pooling_for_Efficient_Deep_Learning_WACV_2023_paper.pdf">WACV</a> 756 757 758 759 Vision HGNN: An Image is More than a Graph of Nodes <br /> 760 <em> </em><br /><a href="https://openaccess.thecvf.com/content/ICCV2023/papers/Han_Vision_HGNN_An_Image_is_More_than_a_Graph_of_ICCV_2023_paper.pdf">ICCV</a> 761 762 763 764 SAL-ViT: Towards Latency Efficient Private Inference on ViT using Selective Attention Search with a Learnable Softmax Approximation <br /> 765 <em> </em><br /><a href="https://openaccess.thecvf.com/content/ICCV2023/papers/Zhang_SAL-ViT_Towards_Latency_Efficient_Private_Inference_on_ViT_using_Selective_ICCV_2023_paper.pdf">ICCV</a> 766 767 768 769 InstaTune: Instantaneous Neural Architecture Search During Fine-Tuning <br /> 770 <em> </em><br /><a href="https://openaccess.thecvf.com/content/ICCV2023W/RCV/papers/Sridhar_InstaTune_Instantaneous_Neural_Architecture_Search_During_Fine-Tuning_ICCVW_2023_paper.pdf">CVF</a> 771 772 773 774 NeRFool: Uncovering the Vulnerability of Generalizable Neural Radiance Fields against Adversarial Perturbations <br /> 775 <em> </em><br /><a href="https://proceedings.mlr.press/v202/fu23g/fu23g.pdf">PMLR</a> 776 777 778 779 Learning to Linearize Deep Neural Networks for Secure and Efficient Private Inference <br /> 780 <em> </em><br /><a href="https://openreview.net/forum?id=BGF9IeDfmlH">OpenReview</a> 781 782 783 784 C2PI: An Efficient Crypto-Clear Two-Party Neural Network Private Inference <br /> 785 <em> </em><br /><a href="https://openreview.net/pdf?id=3ieUGDWGrt4">arxiv</a> 786 787 --> 788 789<h3 style="padding-bottom: 0px;margin-bottom: 0; text-align: center;">2026</h3> 790<hr style="border: 2;height: 2px;width: 90%;" /> 791 792<p>Shubhashis Roy Dipta<em>, Tz-Ying Wu</em>, Subarna Tripathi <br /> 793 <b> Advancing Reference-free Evaluation of Video Captions with Factual Analysis </b> <br /> 794 ACL 2026. <br /> 795 <a href="">[ACL]</a> <a href="">[project]</a> 796 <a href="">[code]</a></p> 797 798<!-- 799 --> 800 801<p>Tz-Ying Wu, Sharath Nittur Sridhar, Subarna Tripathi <br /> 802 <b> Harnessing Object Grounding for Time-Sensitive Video Understanding </b> <br />
803 WACV 2026. <br /> 804 <a href="">[WACV]</a> <a href="">[project]</a> 805 <a href="">[code]</a></p> 806 807<!-- 808 --> 809 810<p>Yuchen Xia, Divyam Sharma, Yichao Yuan, Souvik Kundu, Nishil Talati <br /> 811 <b> MoDM: Efficient Serving for Image Generation via Mixture-of-Diffusion Models </b> <br /> 812 ASPLOS 2026. <br /> 813 <a href="">[ASPLOS]</a> <a href="">[project]</a> 814 <a href="">[code]</a></p> 815 816<!-- 817 --> 818 819<h3 style="padding-bottom: 0px;margin-bottom: 0; text-align: center;">2025</h3> 820<hr style="border: 2;height: 2px;width: 90%;" /> 821 822<p>Ivan Rodin<em>, Tz-Ying Wu</em>, Kyle Min, Sharath Nittur Sridhar, Antonino Furnari, Subarna Tripathi, Giovanni Maria Farinella <br /> 823 <b> EASG-Bench: Video Q&A Benchmark with Egocentric Action Scene Graphs </b> <br /> 824 ICCV Workshop 2025. <br /> 825 <a href="">[ICCVW]</a> <a href="">[project]</a> 826 <a href="">[code]</a></p> 827 828<!-- 829 --> 830 831<p>Rajarshi Roy, Devleena Das, Ankesh Banerjee, Arjya Bhattacharjee, Kousik Dasgupta, Subarna Tripathi <br /> 832 <b> ByDeWay: Boost Your multimodal LLM with DEpth prompting in a training-free Way </b> <br /> 833 ICCV Workshop 2025. <br /> 834 <a href="">[ICCVW]</a> <a href="">[project]</a> 835 <a href="">[code]</a></p> 836 837<!-- 838 --> 839 840<p>Tz-Ying Wu, Tahani Trigui, Sharath Nittur Sridhar, Anand Bodas, Subarna Tripathi <br /> 841 <b> Toward Scalable Video Narration: A Training-free Approach Using Multimodal Large Language Models </b> <br /> 842 ICCV Workshop 2025. <br /> 843 <a href="">[ICCVW]</a> <a href="">[project]</a> 844 <a href="">[code]</a></p> 845 846<!-- 847 --> 848 849<p>Jongseo Lee, Kyungho Bae, Kyle Min, Gyeong-Moon Park, Jinwoo Choi <br /> 850 <b> ESSENTIAL: Episodic and Semantic Memory Integration for Video Class-Incremental Learning </b> <br /> 851 ICCV 2025. <br /> 852 <a href="">[ICCV]</a> <a href="">[project]</a> 853 <a href="">[code]</a></p> 854 855<!-- 856 --> 857 858<p>Akshat Ramachandran, Mingyu Lee, Huan Xu, Souvik Kundu, Tushar Krishna <br /> 859 <b> OuroMamba: A Data-Free Quantization Framework for Vision Mamba Models </b> <br /> 860 ICCV 2025. <br /> 861 <a href="">[ICCV]</a> <a href="">[project]</a> 862 <a href="">[code]</a></p> 863 864<!-- 865 --> 866 867<p>Wenhao Zheng, Yixiao Chen, Weitong Zhang, Souvik Kundu, Yun Li, Zhengzhong Liu, Eric P Xing, Hongyi Wang, Huaxiu Yao <br /> 868 <b> CITER: Collaborative Inference for Efficient Large Language Model Decoding with Token-Level Routing </b> <br /> 869 COLM 2025. <br /> 870 <a href="">[COLM]</a> <a href="">[project]</a> 871 <a href="">[code]</a></p> 872 873<!-- 874 --> 875 876<p>Runjin Chen, Zhenyu Zhang, Junyuan Hong, Souvik Kundu, Zhangyang Wang <br /> 877 <b> SEAL: Steerable reasoning calibration of large language models for free </b> <br /> 878 COLM 2025. <br /> 879 <a href="">[COLM]</a> <a href="">[project]</a> 880 <a href="">[code]</a></p> 881 882<!-- 883 --> 884 885<p>Chenyu Li, Oscar Michel, Xichen Pan, Sainan Liu, Mike Roberts, Saining Xie <br /> 886 <b> PISA Experiments: Exploring Physics Post-Training for Video Diffusion Models by Watching Stuff Drop </b> <br /> 887 ICML 2025. <br /> 888 <a href="">[ICML]</a> <a href="">[project]</a> 889 <a href="">[code]</a></p> 890 891<!-- 892 --> 893 894<p>Yeonju Ro, Zhenyu Zhang, Souvik Kundu, Zhangyang Wang, Aditya Akella <br /> 895 <b> On-the-Fly Adaptive Distillation of Transformer to Dual-State Linear Attention for Long-Context LLM Serving </b> <br /> 896 ICML 2025. <br /> 897 <a href="">[ICML]</a> <a href="">[project]</a> 898 <a href="">[code]</a></p> 899 900<!-- 901 --> 902 903<p>Murthy L, Subarna Tripathi <br /> 904 <b> PALADIN : Robust Neural Fingerprinting for Text-to-Image Diffusion Models </b> <br /> 905 CVPR Workshop 2025. <br /> 906 <a href="">[CVPRW]</a> <a href="">[project]</a> 907 <a href="">[code]</a></p> 908 909<!-- 910 --> 911 912<p>Zhifan Ye, Zheng Wang, Kejing Xia, Jihoon Hong, Leshu Li, Lexington Whalen, Cheng Wan, Yonggan Fu, Celine Lin, Souvik Kundu <br /> 913 <b> LAMB: A Training-Free Method to Enhance the Long-Context Understanding of SSMs via Attention-Guided Token Filtering </b> <br /> 914 ACL 2025. <br /> 915 <a href="">[ACL]</a> <a href="">[project]</a> 916 <a href="">[code]</a></p> 917 918<!-- 919 --> 920 921<p>Akshat Ramachandran, Souvik Kundu, Arnab Raha, Shamik Kundu, Deepak K Mathaikutty, Tushar Krishna <br /> 922 <b> Accelerating LLM Inference with Flexible N: M Sparsity via A Fully Digital Compute-in-Memory Accelerator </b> <br /> 923 DAC 2025. <br /> 924 <a href="">[DAC]</a> <a href="">[project]</a> 925 <a href="">[code]</a></p> 926 927<!-- 928 --> 929 930<p>Souvik Kundu, Anahita B., S Yu, P Howard, Tiep Le, Sharath Nittur Sridhar, H Kang,
930V Lal <br /> 931 <b> LVLM-Compress-Bench: Benchmarking the Broader Impact of Large Vision-Language Model Compression </b> <br /> 932 NAACL 2025. <br /> 933 <a href="">[NAACL]</a> <a href="">[project]</a> 934 <a href="">[code]</a></p> 935 936<!-- 937 --> 938 939<p>Austin Henry Cheng, Alston Lo, Kin Long Kelvin Lee, Santiago Miret, Alan Aspuru-Guzik <br /> 940 <b> Stiefel Flow Matching for Moment-Constrained Structure Elucidation </b> <br /> 941 ICLR 2025. <br /> 942 <a href="">[ICLR]</a> <a href="">[project]</a> 943 <a href="">[code]</a></p> 944 945<!-- 946 --> 947 948<p>Daniel Levy, Siba Smarak Panigrahi, Sékou-Oumar Kaba, Qiang Zhu, Kin Long Kelvin Lee, Mikhail Galkin, Santiago Miret, Siamak Ravanbakhsh <br /> 949 <b> SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Models </b> <br /> 950 ICLR 2025. <br /> 951 <a href="">[ICLR]</a> <a href="">[project]</a> 952 <a href="">[code]</a></p> 953 954<!-- 955 --> 956 957<p>Qianggang Ding, Santiago Miret, Bang Liu <br /> 958 <b> MatExpert: Decomposing Materials Discovery By Mimicking Human Experts </b> <br /> 959 ICLR 2025. <br /> 960 <a href="">[ICLR]</a> <a href="">[project]</a> 961 <a href="">[code]</a></p> 962 963<!-- 964 --> 965 966<p>Yonghao Zhuang, L. Hu, Yun Longfei, Souvik Kundu, Liu Zhengzhong, Eric P. Xing, Hao Zhang <br /> 967 <b> Scaling Long Context Training Data by Long-Distance Referrals </b> <br /> 968 ICLR 2025. <br /> 969 <a href="">[ICLR]</a> <a href="">[project]</a> 970 <a href="">[code]</a></p> 971 972<!-- 973 --> 974 975<p>D. Jang, S. Park, June Y. Yang, Y. Jung, J. Yun, Souvik Kundu, Sung-Yub Kim, Eunho Yang <br /> 976 <b> LANTERN: Accelerating Visual Autoregressive Models with Relaxed Speculative Decoding </b> <br /> 977 ICLR 2025. <br /> 978 <a href="">[ICLR]</a> <a href="">[project]</a> 979 <a href="">[code]</a></p> 980 981<!-- 982 --> 983 984<p>Jianan Zhao, Hesham Mostafa, Mikhail Galkin, Michael M. Bronstein, Zhaocheng Zhu, Jian Tang <br /> 985 <b> GraphAny: A Foundation Model for Node Classification on Any Graph </b> <br /> 986 ICLR 2025. <br /> 987 <a href="">[ICLR]</a> <a href="https://github.com/DeepGraphLearning/GraphAny">[project]</a> 988 <a href="https://github.com/DeepGraphLearning/GraphAny">[code]</a></p> 989 990<!-- 991 --> 992 993<p>Souvik Kundu, Seyedarmin Azizi, M. Sadeghi, Massoud Pedram <br /> 994 <b> MambaExtend: A Training-Free Approach to Improve Long Context Extension of Mamba </b> <br /> 995 ICLR 2025. <br /> 996 <a href="">[ICLR]</a> <a href="">[project]</a> 997 <a href="">[code]</a></p> 998 999<!-- 1000 --> 1001 1002<p>SeyedArmin Azizi, Souvik Kundu, M. Sadeghi, Massoud Pedram <br /> 1003 <b> QMambaExtend: Improving Long-Context Extension of Memory-Efficient Mamba Models </b> <br /> 1004 ICLR workshop 2025. <br /> 1005 <a href="">[ICLR workshop]</a> <a href="">[project]</a> 1006 <a href="">[code]</a></p> 1007 1008<!-- 1009 --> 1010 1011<p>Sihwan Park, Doohyuk Jang, Sung-Yub Kim, Souvik Kundu, Eunho Yang <br /> 1012 <b> LANTERN++: Enhanced Relaxed Speculative Decoding with Static Tree Drafting for Visual Auto-regressive Models </b> <br /> 1013 ICLR workshop 2025. <br /> 1014 <a href="">[ICLR workshop]</a> <a href="">[project]</a> 1015 <a href="">[code]</a></p> 1016 1017<!-- 1018 --> 1019 1020<p>S. Sarkar, G. Datta, Souvik Kundu, K. Zheng, C. Bhattacharyya, P. A. Beerel <br /> 1021 <b> MaskVD: Region Masking for Efficient Video Object Detection </b> <br /> 1022 WACV 2025. <br /> 1023 <a href="">[WACV]</a> <a href="">[project]</a> 1024 <a href="">[code]</a></p> 1025 1026<!-- 1027 --> 1028 1029<p>Tz-Ying Wu, Kyle Min, Subarna Tripathi, Nuno Vasconcelos <br /> 1030 <b> Ego-VPA: Egocentric Video Understanding with Parameter-efficient Adaptation </b> <br /> 1031 WACV 2025. <br /> 1032 <a href="">[WACV]</a> <a href="">[project]</a> 1033 <a href="">[code]</a></p> 1034 1035<!-- 1036 --> 1037 1038<p>Utkarsh Nath, Rajeev Goel, Kyle Min, Yezhou Yang, Eun Som Jeon <br /> 1039 <b> Deep Geometric Moments Promote Shape Consistency in Text-to-3D Generation </b> <br /> 1040 WACV 2025. <br /> 1041 <a href="">[WACV]</a> <a href="">[project]</a> 1042 <a href="">[code]</a></p> 1043 1044<!-- 1045 --> 1046 1047<p>Shreya Saha, Lucas Liang, Shan Lin, Jingpei Lu, Michael Yip, Sainan Liu <br /> 1048 <b> BASED: Bundle-Adjusting Surgical Endoscopic Dynamic Video Reconstruction using Neural Radiance Fields </b> <br /> 1049 WACV 2025. <br /> 1050 <a href="">[WACV]</a> <a href="">[project]</a> 1051 <a href="">[code]</a></p> 1052 1053<!-- 1054 --> 1055 1056<h3 style="padding-bottom: 0px;margin-bottom: 0; text-align: center;">2024</h3> 1057<hr style="border: 2;height: 2px;width: 90%;" /> 1058 1059<p>Anthony Sarah, Sharath Nittur Sridhar, Maciej Szankin, Sairam Sundaresan <br /> 1060 <b> LLaMA-NAS: Efficient Neural Architecture Search for Large Language Models </b> <br /> 1061 ECCV 2024. <br /> 1062 <a href="https://arxiv.org/abs/2405.18377">[ECCV]</a> <a href="">[project]</a> 1063 <a href="">[code]</a></p> 1064 1065<!-- 1066 <a href="https://arxiv.org/abs/2405.18377">[ECCV]</a> 1067 1068 --> 1069 1070<p>
1070Shamik Kundu, Sanjay Das, Sayar Karmakar, Arnab Raha, Souvik Kundu, Yiorgos Makris, Kanad Basu <br /> 1071 <b> Bit-by-Bit: Investigating the Vulnerabilities of Binary Neural Networks to Adversarial Bit Flipping </b> <br /> 1072 TMLR 2024. <br /> 1073 <a href="">[TMLR]</a> <a href="">[project]</a> 1074 <a href="">[code]</a></p> 1075 1076<!-- 1077 --> 1078 1079<p>Changhoon Kim<em>, Kyle Min</em>, Yezhou Yang <br /> 1080 <b> R.A.C.E.: Robust Adversarial Concept Erasure for Secure Text-to-Image Diffusion Model </b> <br /> 1081 ECCV 2024. <br /> 1082 <a href="">[ECCV]</a> <a href="">[project]</a> 1083 <a href="">[code]</a></p> 1084 1085<!-- 1086 --> 1087 1088<p>A. Ramachandran, Souvik Kundu, Tushar Krishna <br /> 1089 <b> CLAMP-ViT: Contrastive Data-Free Learning for Adaptive Post-Training Quantization of ViTs </b> <br /> 1090 ECCV 2024. <br /> 1091 <a href="">[ECCV]</a> <a href="">[project]</a> 1092 <a href="">[code]</a></p> 1093 1094<!-- 1095 --> 1096 1097<p>A. Ramachandran, Souvik Kundu, Tushar Krishna <br /> 1098 <b> GenQ: Quantization in Low Data Regimes with Generative Synthetic Data </b> <br /> 1099 ECCV 2024. <br /> 1100 <a href="https://arxiv.org/pdf/2312.05272">[ECCV]</a> <a href="">[project]</a> 1101 <a href="">[code]</a></p> 1102 1103<!-- 1104 <a href="https://arxiv.org/pdf/2312.05272">[ECCV]</a> 1105 1106 --> 1107 1108<p>Souvik Kundu, Sharath Nittur Sridhar, Maciej Szankin, Sairam Sundaresan <br /> 1109 <b> Sensi-BERT: Towards Sensitivity Driven Fine-Tuning for Parameter Efficient BERT </b> <br /> 1110 ICASSP 2024. <br /> 1111 <a href="https://arxiv.org/abs/2307.11764">[ICASSP]</a> <a href="">[project]</a> 1112 <a href="">[code]</a></p> 1113 1114<!-- 1115 <a href="https://arxiv.org/abs/2307.11764">[ICASSP]</a> 1116 1117 --> 1118 1119<p>Sharath Nittur Sridhar, Maciej Szankin, Fang Chen, Sairam Sundaresan, Anthony Sarah <br /> 1120 <b> SimQ-NAS: Simultaneous Quantization Policy and Neural Architecture Search </b> <br /> 1121 Association for the Advancement of Artificial Intelligence (AAAI) 2024. <br /> 1122 <a href="https://arxiv.org/abs/2312.13301">[arXiv]</a> <a href="">[project]</a> 1123 <a href="">[code]</a></p> 1124 1125<!-- 1126 <a href="https://arxiv.org/abs/2312.13301">[arXiv]</a> 1127 1128 --> 1129 1130<p>Raj Ghugare, Santiago Miret, Adriana Hugessen, Mariano Phielipp, Glen Berseth <br /> 1131 <b> Searching the Space of High-Value Molecules Using Reinforcement Learning and Language Models </b> <br /> 1132 ICLR 2024. <br /> 1133 <a href="">[ICLR]</a> <a href="">[project]</a> 1134 <a href="">[code]</a></p> 1135 1136<!-- 1137 --> 1138 1139<p>Mikhail Galkin, Xinyu Yuan, Hesham Mostafa, Jian Tang, Zhaocheng Zhu <br /> 1140 <b> Towards Foundation Models for Knowledge Graph Reasoning </b> <br /> 1141 ICLR 2024. <br /> 1142 <a href="">[ICLR]</a> <a href="">[project]</a> 1143 <a href="">[code]</a></p> 1144 1145<!-- 1146 --> 1147 1148<p>Hongyi Wang, Felipe Maia Polo, Yuekai Sun, Souvik Kundu, Eric Xing, Mikhail Yurochkin <br /> 1149 <b> Fusing Models with Complementary Expertise </b> <br /> 1150 ICLR 2024. <br /> 1151 <a href="https://openreview.net/pdf?id=PhMrGCMIRL">[ICLR]</a> <a href="">[project]</a> 1152 <a href="">[code]</a></p> 1153 1154<!-- 1155 <a href="https://openreview.net/pdf?id=PhMrGCMIRL">[ICLR]</a> 1156 1157 --> 1158 1159<p>Souvik Kundu, Zeyu Liu, Anni Li, Junrui Wan, Lianghao Jiang, Peter A. Beerel <br /> 1160 <b> AFLoRA: Adaptive Freezing of Low-Rank Adapters for Parameter Efficient Fine-Tuning of Language Models </b> <br /> 1161 ACL (best paper nomination from Meta-reviewer) 2024. <br /> 1162 <a href="https://arxiv.org/pdf/2403.13269.pdf">[ACL]</a> <a href="">[project]</a> 1163 <a href="">[code]</a></p> 1164 1165<!-- 1166 <a href="https://arxiv.org/pdf/2403.13269.pdf">[ACL]</a> 1167 1168 --> 1169 1170<p>Souvik Kundu, Sreetama Sarkar, Peter A. Beerel <br /> 1171 <b> RLNet: Robust Linearized Networks for Efficient Private Inference </b> <br /> 1172 CVPR (Trusted Computer Vision) 2024. <br /> 1173 <a href="">[CVPRW]</a> <a href="">[project]</a> 1174 <a href="https://github.com/sreetamasarkar/rlnet/tree/main">[code]</a></p> 1175 1176<!-- 1177 --> 1178 1179<p>Sreetama Sarkar, Souvik Kundu, Kai Zheng, Peter A. Beerel <br /> 1180 <b> Block Selective Reprogramming for On-device Training of Vision Transformers </b> <br /> 1181 CVPR (ECV) ORAL presentation 2024. <br /> 1182 <a href="">[CVPRW]</a> <a href="">[project]</a> 1183 <a href="https://github.com/sreetamasarkar/BSR">[code]</a></p> 1184 1185<!-- 1186 --> 1187 1188<p>Jose Rojas Chaves, Subarna Tripathi <br /> 1189 <b> VideoSAGE: Video Summarization with Graph Representation Learning </b> <br /> 1190 CVPR 2024. <br /> 1191 <a href="https://arxiv.org/pdf/2404.10539.pdf">[CVPRW]</a> <a href="">[project]</a> 1192 <a href="">[code]</a></p> 1193 1194<!-- 1195 <a href="https://arxiv.org/pdf/2404.10539.pdf">[CVPRW]</a> 1196 1197 --> 1198 1199<p>Hengyue Liu, Kyle Min, Hector A Valdez, Subarna Tripathi <br /> 1200 <b> LAVITI: Contrastive Language Video Time Pre-training </b> <br /> 1201 CVPR (EgoVis) 2024. <br /> 1202 <a href="">[CVPRW]</a> <a href="">[project]</a> 1203 <a href="">[code]</a></p> 1204 1205<!-- 1206 --> 1207 1208<p>Lu Yin, Shiwei Liu, Ajay Jaiswal, Souvik Kundu, Zhangyang (Atlas) Wang <br /> 1209 <b> Pruning Small Magnitude Weights Monotonically and Irreversibly Impairs Difficult Downstream Tasks in LLMs </b> <br /> 1210 ICML 2024. <br /> 1211 <a href="">[arXiv]</a> <a href="">[project]</a> 1212 <a href="">[code]</a></p> 1213 1214<!-- 1215 --> 1216 1217<p>Souvik Kundu, Anthony Sarah, Vinay Joshi, Om J Omer, Sreenivas Subramoney <br /> 1218 <b> Towards Joint Optimization for DNN Architecture and Configuration for Compute-In-Memory Hardware </b> <br /> 1219 TinyML 2024. <br /> 1220 <a href="https://arxiv.org/abs/2402.11780">[arXiv]</a> <a href="">[project]</a> 1221 <a href="">[code]</a></p> 1222 1223<!-- 1224 <a href="https://arxiv.org/abs/2402.11780">[arXiv]</a> 1225 1226 --> 1227 1228<p>Tz-Ying Wu, Kyle Min, Subarna Tripathi, Nuno Vasconcelos <br /> 1229 <b> Ego-VPA: Egocentric Video Understanding with Parameter-efficient Adaptation </b> <br /> 1230 under review 2024. <br /> 1231 <a href="">[Under review]</a> <a href="">[project]</a> 1232 <a href="">[code]</a></p> 1233 1234<!-- 1235 --> 1236 1237<p>Ting Hsuan Liao, Sainan Liu, Ilke Demir, Jia-Bin Huang <br /> 1238 <b> PASTE: Prompt-Aligned Scene TExture synthesis </b> <br /> 1239 under review 2024. <br /> 1240 <a href="">[Under review]</a> <a href="">[project]</a> 1241 <a href="">[code]</a></p> 1242 1243<!-- 1244 --> 1245 1246<p>Jiraphon Yenphraphai, Xichen Pan, Sainan Liu, Daniele Panozzo, Saining Xie <br /> 1247 <b> Image Sculpting: Precise Object Editing with 3D Geometry Control </b> <br /> 1248 CVPR 2024. <br /> 1249 <a href="https://arxiv.org/abs/2401.01702">[CVPR]</a> <a href="https://image-sculpting.github.io/">[project]</a> 1250 <a href="https://github.com/vision-x-nyu/image-sculpting">[code]</a></p> 1251 1252<!-- 1253 <a href="https://arxiv.org/abs/2401.01702">[CVPR]</a> 1254 1255 --> 1256 1257<p>Changhoon Kim, Kyle Min, Maitreya Patel, Sheng Cheng, Yezhou Yang <br /> 1258 <b> WOUAF: Weight Modulation for User Attribution and Fingerprinting in Text-to-Image Diffusion Models </b> <br /> 1259 CVPR 2024. <br /> 1260 <a href="https://arxiv.org/abs/2306.04744">[CVPR]</a> <a href="https://wouaf.vercel.app/">[project]</a>
1261 <a href="https://huggingface.co/spaces/wouaf/WOUAF-Text-to-Image">[code]</a></p> 1262 1263<!-- 1264 <a href="https://arxiv.org/abs/2306.04744">[CVPR]</a> 1265 1266 --> 1267 1268<p>Ivan Rodin, Antonino Furnari, Kyle Min, Subarna Tripathi, Giovanni Maria Farinella <br /> 1269 <b> Action Scene Graphs for Long-Form Understanding of Egocentric Videos </b> <br /> 1270 CVPR 2024. <br /> 1271 <a href="https://arxiv.org/abs/2312.03391">[CVPR]</a> <a href="">[project]</a> 1272 <a href="https://github.com/fpv-iplab/EASG">[code]</a></p> 1273 1274<!-- 1275 <a href="https://arxiv.org/abs/2312.03391">[CVPR]</a> 1276 1277 --> 1278 1279<p>Sainan Liu, Shan Lin, Jingpei Lu, Shreya Saha, Alexey Supikov, Michael Yip <br /> 1280 <b> BAA-NGP: Bundle-Adjusting Accelerated Neural Graphics Primitives </b> <br /> 1281 CVPRW 2024. <br /> 1282 <a href="https://arxiv.org/abs/2306.04166">[ArXiv]</a> <a href="">[project]</a> 1283 <a href="https://github.com/IntelLabs/baa-ngp">[code]</a></p> 1284 1285<!-- 1286 <a href="https://arxiv.org/abs/2306.04166">[ArXiv]</a> 1287 1288 --> 1289 1290<h3 style="padding-bottom: 0px;margin-bottom: 0; text-align: center;">2023</h3> 1291<hr style="border: 2;height: 2px;width: 90%;" /> 1292 1293<p>Shreya Saha, Sainan Liu, Shan Lin, Jingpei Lu, Michael Yip <br /> 1294 <b> BASED: Bundle-Adjusting Surgical Endoscopic Dynamic Video Reconstruction using Neural Radiance Fields </b> <br /> 1295 arXiv 2023. <br /> 1296 <a href="https://arxiv.org/abs/2309.15329">[ArXiv]</a> <a href="https://sites.google.com/ucsd.edu/based-arclab-2023/">[project]</a> 1297 <a href="">[code]</a></p> 1298 1299<!-- 1300 <a href="https://arxiv.org/abs/2309.15329">[ArXiv]</a> 1301 1302 --> 1303 1304<p>Kyle Min <br /> 1305 <b> STHG: Spatial-Temporal Heterogeneous Graph Learning for Advanced Audio-Visual Diarization </b> <br /> 1306 Ego4D challenge at CVPR 2023. <br /> 1307 <a href="https://arxiv.org/abs/2306.10608">[ArXiv]</a> <a href="">[project]</a> 1308 <a href="">[code]</a></p> 1309 1310<!-- 1311 <a href="https://arxiv.org/abs/2306.10608">[ArXiv]</a> 1312 1313 --> 1314 1315<p>Yi Li, Kyle Min, Subarna Tripathi, Nuno Vasconcelos <br /> 1316 <b> SViTT: Temporal Learning of Sparse Video-Text Transformer </b> <br /> 1317 Computer Vision and Pattern Recognition (CVPR), 2023. <br /> 1318 <a href="https://arxiv.org/abs/2304.08809">[arXiv]</a> <a href="http://svcl.ucsd.edu/projects/svitt/">[project]</a> 1319 <a href="">[code]</a></p> 1320 1321<!-- 1322 <a href="https://arxiv.org/abs/2304.08809">[arXiv]</a> 1323 1324 --> 1325 1326<p>Sayak Nag, Kyle Min, Subarna Tripathi, Amit Roy Choudhury <br /> 1327 <b> Unbiased video scene graph generation </b> <br /> 1328 Computer Vision and Pattern Recognition (CVPR), 2023. <br /> 1329 <a href="https://arxiv.org/abs/2304.00733">[arXiv]</a> <a href="">[project]</a> 1330 <a href="https://github.com/sayaknag/unbiasedSGG.git">[code]</a></p> 1331 1332<!-- 1333 <a href="https://arxiv.org/abs/2304.00733">[arXiv]</a> 1334 1335 --> 1336 1337<p>Shengyu Feng, Hesham Mostafa, Marcel Nassar, Somdeb Majumdar, Subarna Tripathi <br /> 1338 <b> Exploiting Long-Term Dependencies for Generating Dynamic Scene Graphs </b> <br /> 1339 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2023. <br /> 1340 <a href="https://openaccess.thecvf.com/content/WACV2023/papers/Feng_Exploiting_Long-Term_Dependencies_for_Generating_Dynamic_Scene_Graphs_WACV_2023_paper.pdf">[CVF]</a> <a href="">[project]</a> 1341 <a href="https://github.com/Shengyu-Feng/DSG-DETR">[code]</a></p> 1342 1343<!-- 1344 <a href="https://openaccess.thecvf.com/content/WACV2023/papers/Feng_Exploiting_Long-Term_Dependencies_for_Generating_Dynamic_Scene_Graphs_WACV_2023_paper.pdf">[CVF]</a> 1345 1346 --> 1347 1348<p>Fang Chen, Gourav Datta, Souvik Kundu, Peter A. Beerel <br /> 1349 <b> Self-Attentive Pooling for Efficient Deep Learning </b> <br /> 1350 WACV 2023. <br /> 1351 <a href="https://openaccess.thecvf.com/content/WACV2023/papers/Chen_Self-Attentive_Pooling_for_Efficient_Deep_Learning_WACV_2023_paper.pdf">[WACV]</a> <a href="">[project]</a> 1352 <a href="">[]</a></p> 1353 1354<!-- 1355 <a href="https://openaccess.thecvf.com/content/WACV2023/papers/Chen_Self-Attentive_Pooling_for_Efficient_Deep_Learning_WACV_2023_paper.pdf">[WACV]</a> 1356 1357 --> 1358 1359<p>Yan Han, Peihao Wang, <strong>Souvik Kundu</strong>, Ding Ying, Zhangyang (Atlas) Wang <br /> 1360 <b> Vision HGNN: An Image is More than a Graph of Nodes </b> <br /> 1361 International Conference on Computer Vision (ICCV) ORAL, 2023. <br /> 1362 <a href="https://openaccess.thecvf.com/content/ICCV2023/papers/Han_Vision_HGNN_An_Image_is_More_than_a_Graph_of_ICCV_2023_paper.pdf">[ICCV]</a> <a href="">[project]</a> 1363 <a href="https://github.com/VITA-Group/ViHGNN">[]</a></p> 1364 1365<!-- 1366 <a href="https://openaccess.thecvf.com/content/ICCV2023/papers/Han_Vision_HGNN_An_Image_is_More_than_a_Graph_of_ICCV_2023_paper.pdf">[ICCV]</a> 1367 1368 --> 1369 1370<p>Souvik Kundu, Yuke Zhang, Dake Chen*, Chenghao Li, Peter A. Beerel <br /> 1371 <b> SAL-ViT: Towards Latency Efficient Private Inference on ViT using Selective Attention Search with a Learnable Softmax Approximation </b> <br /> 1372 International Conference on Computer Vision (ICCV), 2023. <br /> 1373 <a href="https://openaccess.thecvf.com/content/ICCV2023/papers/Zhang_SAL-ViT_Towards_Latency_Efficient_Private_Inference_on_ViT_using_Selective_ICCV_2023_paper.pdf">[ICCV]</a> <a href="">[project]</a> 1374 <a href="">[]</a></p> 1375 1376<!-- 1377 <a href="https://openaccess.thecvf.com/content/ICCV2023/papers/Zhang_SAL-ViT_Towards_Latency_Efficient_Private_Inference_on_ViT_using_Selective_ICCV_2023_paper.pdf">[ICCV]</a> 1378 1379 --> 1380 1381<p>
1381Sharath Nittur Sridhar, Souvik Kundu, Sairam Sundaresan, Maciej Szankin, Anthony Sarah <br /> 1382 <b> InstaTune: Instantaneous Neural Architecture Search During Fine-Tuning </b> <br /> 1383 International Conference on Computer Vision (ICCV) Workshop 2023. <br /> 1384 <a href="https://openaccess.thecvf.com/content/ICCV2023W/RCV/papers/Sridhar_InstaTune_Instantaneous_Neural_Architecture_Search_During_Fine-Tuning_ICCVW_2023_paper.pdf">[CVF]</a> <a href="">[project]</a> 1385 <a href="">[]</a></p> 1386 1387<!-- 1388 <a href="https://openaccess.thecvf.com/content/ICCV2023W/RCV/papers/Sridhar_InstaTune_Instantaneous_Neural_Architecture_Search_During_Fine-Tuning_ICCVW_2023_paper.pdf">[CVF]</a> 1389 1390 --> 1391 1392<p>Yonggan Fu, Ye Yuan, Souvik Kundu, Shang Wu, Shunyao Zhang, Celine Lin <br /> 1393 <b> NeRFool: Uncovering the Vulnerability of Generalizable Neural Radiance Fields against Adversarial Perturbations </b> <br /> 1394 International Conference on Machine Learning (ICML), 2023. <br /> 1395 <a href="https://proceedings.mlr.press/v202/fu23g/fu23g.pdf">[PMLR]</a> <a href="">[project]</a> 1396 <a href="https://github.com/GATECH-EIC/NeRFool">[]</a></p> 1397 1398<!-- 1399 <a href="https://proceedings.mlr.press/v202/fu23g/fu23g.pdf">[PMLR]</a> 1400 1401 --> 1402 1403<p>Souvik Kundu, Shunlin Lu, Yuke Zhang, Jacqueline Liu, Peter A. Beerel <br /> 1404 <b> Learning to Linearize Deep Neural Networks for Secure and Efficient Private Inference </b> <br /> 1405 International Conference on Learning Representation (ICLR), 2023. <br /> 1406 <a href="https://openreview.net/forum?id=BGF9IeDfmlH">[OpenReview]</a> <a href="">[project]</a> 1407 <a href="">[]</a></p> 1408 1409<!-- 1410 <a href="https://openreview.net/forum?id=BGF9IeDfmlH">[OpenReview]</a> 1411 1412 --> 1413 1414<p>Yuke Zhang, Dake Chen, Souvik Kundu, Haomei Liu, Ruiheng Peng, Peter A. Beerel <br /> 1415 <b> C2PI: An Efficient Crypto-Clear Two-Party Neural Network Private Inference </b> <br /> 1416 Design Automation Conference (DAC), 2023. <br /> 1417 <a href="https://openreview.net/pdf?id=3ieUGDWGrt4">[arxiv]</a> <a href="">[project]</a> 1418 <a href="">[]</a></p> 1419 1420<!-- 1421 <a href="https://openreview.net/pdf?id=3ieUGDWGrt4">[arxiv]</a> 1422 1423 --> 1424 1425<h3 style="padding-bottom: 0px;margin-bottom: 0; text-align: center;">2022</h3> 1426<hr style="border: 2;height: 2px;width: 90%;" /> 1427 1428<p>Shaowei Liu, Subarna Tripathi, Somdeb Majumdar, Xiaolong Wang <br /> 1429 <b> Joint Hand Motion and Interaction Hotspots Prediction from Egocentric Videos </b> <br /> 1430 Computer Vision and Pattern Recognition (CVPR), 2022. <br /> 1431 <a href="https://arxiv.org/abs/2204.01696">[arXiv]</a> <a href="https://stevenlsw.github.io/hoi-forecast/">[project]</a> 1432 <a href="https://github.com/stevenlsw/hoi-forecast">[code]</a></p> 1433 1434<!-- 1435 <a href="https://arxiv.org/abs/2204.01696">[arXiv]</a> 1436 1437 --> 1438 1439<p>Xiang Zhang, Yongwen Su, Subarna Tripathi, Zhuowen Tu <br /> 1440 <b> Text Spotting Transformers </b> <br /> 1441 Computer Vision and Pattern Recognition (CVPR), 2022. <br /> 1442 <a href="https://arxiv.org/abs/2204.01918">[arXiv]</a> <a href="">[project]</a> 1443 <a href="https://github.com/mlpc-ucsd/TESTR">[code]</a></p> 1444 1445<!-- 1446 <a href="https://arxiv.org/abs/2204.01918">[arXiv]</a> 1447 1448 --> 1449 1450<p>Kyle Min<em>, Sourya Roy</em>, Subarna Tripathi, Tanaya Guha, Somdeb Majumdar <br /> 1451 <b> Learning long-term spatial-temporal graphs for active speaker detection </b> <br /> 1452 European Conference on Computer Vision (ECCV), 2022. <br /> 1453 <a href="https://arxiv.org/pdf/2207.07783.pdf">[arXiv]</a> <a href="">[project]</a> 1454 <a href="https://github.com/SRA2/SPELL">[code]</a></p> 1455 1456<!-- 1457 <a href="https://arxiv.org/pdf/2207.07783.pdf">[arXiv]</a> 1458 1459 --> 1460 1461<p>Alakh Desai, Tz-Ying Wu, Subarna Tripathi, Nuno Vasconcelos <br /> 1462 <b> Single-Stage Visual Relationship Learning using Conditional Queries </b> <br /> 1463 Neural Information Processing Systems (NeurIPS), 2022. <br /> 1464 <a href="https://openreview.net/forum?id=hPVXHzzK0z">[openReview]</a> <a href="">[project]</a> 1465 <a href="">[code]</a></p> 1466 1467<!-- 1468 <a href="https://openreview.net/forum?id=hPVXHzzK0z">[openReview]</a> 1469 1470 --> 1471 1472<p>Sharath Nittur Sridhar, Anthony Sarah, Sairam Sundaresan <br /> 1473 <b> TrimBERT: Tailoring BERT for Trade-offs </b> <br /> 1474 International Conference on Machine Learning (ICML) 2022. <br /> 1475 <a href="https://arxiv.org/abs/2202.12411">[arXiv]</a> <a href="">[project]</a> 1476 <a href="">[code]</a></p> 1477 1478<!-- 1479 <a href="https://arxiv.org/abs/2202.12411">[arXiv]</a> 1480 1481 --> 1482 1483<p>Daniel Cummings, Sharath Nittur Sridhar, Anthony Sarah, Maciej Szankin <br /> 1484 <b> Accelerating Neural Architecture Exploration Across Modalities Using Genetic Algorithms </b> <br /> 1485 Genetic and Evolutionary Computation Conference (GECCO) 2022. <br /> 1486 <a href="https://arxiv.org/abs/2202.12934">[arXiv]</a> <a href="">[project]</a> 1487 <a href="">[]</a></p> 1488 1489<!-- 1490 <a href="https://arxiv.org/abs/2202.12934">[arXiv]</a> 1491 1492 --> 1493 1494<p>Daniel Cummings, Anthony Sarah, Sharath Nittur Sridhar, Maciej Szankin, Pablo Munoz, Sairam Sundaresan <br /> 1495 <b> A Hardware-Aware Framework for Accelerating Neural Architecture Search Across Modalities </b> <br /> 1496 Automated Machine Learning (AutoML) 2022. <br /> 1497 <a href="https://arxiv.org/abs/2205.10358">[arXiv]</a> <a href="">[project]</a> 1498 <a href="">[]</a></p> 1499 1500<!-- 1501 <a href="https://arxiv.org/abs/2205.10358">[arXiv]</a> 1502 1503 --> 1504 1505<p>J. Pablo Munoz, Nikolay Lyalyushkin, Chaunté W. Lacewell, Anastasia Senina, Daniel Cummings, Anthony Sarah, Alexander Kozlov, Nilesh Jain <br /> 1506 <b> Automated Super-Network Generation for Scalable Neural Architecture Search </b> <br /> 1507 Automated Machine Learning (AutoML) 2022. <br /> 1508 <a href="https://openreview.net/forum?id=HK-zmbTB8gq">[OpenReview]</a> <a href="">[project]</a> 1509 <a href="">[]</a></p> 1510 1511<!-- 1512 <a href="https://openreview.net/forum?id=HK-zmbTB8gq">[OpenReview]</a> 1513 1514 --> 1515 1516<p>Souvik Kundu, Sairam Sundaresan, M. Pedram, P. A. Beerel <br /> 1517 <b> FLOAT: Fast Learnable Once-for-All Adversarial Training for Tunable Trade-off between Accuracy and Robustness </b> <br /> 1518 ICLR 2022. <br /> 1519 <a href="https://openreview.net/forum?id=MXrIVw-F_a4">[OpenReview]</a> <a href="">[project]</a> 1520 <a href="">[]</a></p> 1521 1522<!-- 1523 <a href="https://openreview.net/forum?id=MXrIVw-F_a4">[OpenReview]</a> 1524 1525 --> 1526 1527<p>Souvik Kundu, Sairam Sundaresan, Sharath Nittur Sridhar, S Lu, H Tang, P. A. Beerel <br /> 1528 <b> Sparse Mixture Once-for-all Adversarial Training for Efficient in-situ Trade-off between Accuracy and Robustness of DNNs </b> <br /> 1529 International Conference on Acoustics, Speech, and Signal Processing (ICASSP) 2022. <br /> 1530 <a href="https://arxiv.org/abs/2302.03523">[arXiv]</a> <a href="">[project]</a> 1531 <a href="">[]</a></p> 1532 1533<!-- 1534 <a href="https://arxiv.org/abs/2302.03523">[arXiv]</a> 1535 1536 --> 1537 1538<h3 style="padding-bottom: 0px;margin-bottom: 0; text-align: center;">2021</h3> 1539<hr style="border: 2;height: 2px;width: 90%;" /> 1540 1541<p>Souvik Kundu, Sairam Sundaresan <br /> 1542 <b> AttentionLite: Towards Efficient Self-attention models for Vision </b> <br /> 1543 International Conference on Acoustics, Speech, and Signal Processing (ICASSP) 2021. <br /> 1544 <a href="https://arxiv.org/abs/2101.05216">[arXiv]</a> <a href="">[project]</a> 1545 <a href="">[]</a></p> 1546 1547<!-- 1548 <a href="https://arxiv.org/abs/2101.05216">[arXiv]</a> 1549 1550 --> 1551 1552<p>Alicja Kwasniewska, Maciej Szankin, Jacek Ruminski, Anthony Sarah, David Gamba <br /> 1553 <b> Improving Accuracy of Respiratory Rate Estimation by Restoring High Resolution Features With Transformers and Recursive Convolutional Models </b> <br /> 1554 CVPRW 2021. <br /> 1555 <a href="https://openaccess.thecvf.com/content/CVPR2021W/CVPM/html/Kwasniewska_Improving_Accuracy_of_Respiratory_Rate_Estimation_by_Restoring_High_Resolution_CVPRW_2021_paper.html">[CVF]</a> <a href="">[project]</a> 1556 <a href="">[]</a></p> 1557 1558<!-- 1559 <a href="https://openaccess.thecvf.com/content/CVPR2021W/CVPM/html/Kwasniewska_Improving_Accuracy_of_Respiratory_Rate_Estimation_by_Restoring_High_Resolution_CVPRW_2021_paper.html">[CVF]</a> 1560 1561 --> 1562 1563<p>
1563Sharath Nittur Sridhar, Anthony Sarah <br /> 1564 <b> Undivided Attention: Are Intermediate Layers Necessary for BERT? </b> <br /> 1565 NeurIPS 2021. <br /> 1566 <a href="https://neurips2021-nlp.github.io/papers/33/CameraReady/Undivided_Attention_Neurips_Workshop_2021_CameraReady.pdf">[NeurIPS]</a> <a href="">[project]</a> 1567 <a href="">[]</a></p> 1568 1569<!-- 1570 <a href="https://neurips2021-nlp.github.io/papers/33/CameraReady/Undivided_Attention_Neurips_Workshop_2021_CameraReady.pdf">[NeurIPS]</a> 1571 1572 --> 1573 1574<p>Hala Lamdouar, Sairam Sundaresan, Anna Jungbluth, Sudeshna Boro Saikia, Amanda Joy Camarata, Nathan Miles, Marcella Scoczynski, Mavis Stone, Anthony Sarah, Andres Munoz-Jaramillo, Ayris Narock, Adam Szabo <br /> 1575 <b> Deep-SWIM: A few-shot learning approach to classify Solar WInd Magnetic field structures </b> <br /> 1576 NeurIPS 2021. <br /> 1577 <a href="https://ml4physicalsciences.github.io/2021/files/NeurIPS_ML4PS_2021_138.pdf">[NeurIPS]</a> <a href="">[project]</a> 1578 <a href="">[]</a></p> 1579 1580<!-- 1581 <a href="https://ml4physicalsciences.github.io/2021/files/NeurIPS_ML4PS_2021_138.pdf">[NeurIPS]</a> 1582 1583 --> 1584 1585<h3 style="padding-bottom: 0px;margin-bottom: 0; text-align: center;">2020</h3> 1586<hr style="border: 2;height: 2px;width: 90%;" /> 1587 1588 1589</div> 1590 1591 </div> 1592 </div> 1593 1594 <div id="footer" class="panel"> 1595 <div class="panel-footer"> 1596 <div class="container-fluid"> 1597 <div class="row"> 1598 <div class="col-sm-4"> 1599 1600 <p>© 2026 Intel Multimodal AI Innovation Lab. Site made with <a href="https://jekyllrb.com">Jekyll</a>; <a href="/aboutwebsite.html"> and modified from <a href="https://www.allanlab.org/"> Allan Lab </a></p> 1601 <p>We are part of the Intelligent Systems Research Lab at <a href="https://www.intel.com/content/www/us/en/research/overview.html">Intel Labs</a>, <a href="https://www.intel.com"> Intel Corporation </a></p> 1602 1603 1604 <p> </p><p> 1605 1606 1607 </div> 1608 <div class="col-sm-4"> 1609 <!-- Contact:<br /> --> 1610 <!-- Intel AI Lab --> 1611 <img src="/images/logopic/IL.png" style="width: 120px"> 1612 <br /> 1613 </div> 1614 1615 1616 <div class="col-sm-4"> 1617 <!-- Contact:<br /> --> 1618 Hello from <br /> 1619 <img src="/images/logopic/ailab.png" style="width: 160px"> 1620 <br /> 1621 </div> 1622 </div> 1623 </div> 1624 </div> 1625</div> 1626
1627<script src="https://ajax.googleapis.com/ajax/libs/jquery/1.11.3/jquery.min.js"></script>
vendor: 1 bytes, line 1627
1627
1628<script src="/js/bootstrap.min.js"></script>
1628 1629 1630 1631 </body> 1632 1633</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.