Bangladesh University of Engineering and Technology

6 related articles

Subscribe to updates

bioRxiv Nov 6, 2026
Flip the strand and DNA language models change their answer; MirrorDNA cuts both strands at mirrored points to give one prediction
DNA language models usually tokenize, train and test on one strand, so a sequence and its reverse complement can get different predictions. MirrorDNA uses a symmetric router that cuts both strands at mirrored positions and proves the whole model is equivariant to the reverse complement. On 12 strand-symmetric Nucleotide Transformer tasks it reaches an MCC of 0.491 on both strands, against 0.152, 0.187 and 0.080 for NT-v2, DNABERT-2 and HyenaDNA.
DNA language models usually tokenize, train and test on one strand, so a sequence and its reverse complement can get different predictions. MirrorDNA uses a symmetric router that cuts both strands at mirrored positions and proves the whole model is equivariant to the reverse complement. On 12 strand-symmetric Nucleotide Transformer tasks it reaches an MCC of 0.491 on both strands, against 0.152, 0.187 and 0.080 for NT-v2, DNABERT-2 and HyenaDNA.
BRAC University Bangladesh University of Engineering and Technology
Nirjhor Datta（First author）M. Sohel Rahman（Corresponding author）
MirrorDNA Reverse-Complement Equivariance DNA Foundation Models
researchsquare.com Oct 22, 2026
SRAM sign-off stops simulating every corner: transferred design points plus conformal calibration still deliver 99.88% yield on a 1-Mb macro
SRAM yield sign-off needs per-cell failure probabilities of 10⁻⁶ to 10⁻¹² at every PVT condition, and running importance sampling at each one is expensive. The method transfers each failure lobe's design point across PVT, corrects it with one SPICE run per lobe, and adds split-conformal margins computed from importance sampling on a random subset of sign-off conditions, escalating undecided conditions to importance sampling. Across all 242 unseen conditions of a SkyWater 130-nm cell and an ASAP7 7-nm cell, no calibrated decision was unsafe, the minimum operating voltage read from the conformal upper limit lay −1 to +34 mV from importance sampling, and a 1-Mb macro delivered at least 99.88% yield.
SRAM yield sign-off needs per-cell failure probabilities of 10⁻⁶ to 10⁻¹² at every PVT condition, and running importance sampling at each one is expensive. The method transfers each failure lobe's design point across PVT, corrects it with one SPICE run per lobe, and adds split-conformal margins computed from importance sampling on a random subset of sign-off conditions, escalating undecided conditions to importance sampling. Across all 242 unseen conditions of a SkyWater 130-nm cell and an ASAP7 7-nm cell, no calibrated decision was unsafe, the minimum operating voltage read from the conformal upper limit lay −1 to +34 mV from importance sampling, and a 1-Mb macro delivered at least 99.88% yield.
Bangladesh University of Engineering and Technology
Nabanita Sarker（First author, Corresponding author）
Design-Point Transfer Conformal Prediction SRAM Reliability Sign-Off First-Order Reliability Method (FORM)Importance Sampling
arXiv Oct 14, 2026
Label theft in vertical federated learning now leaves a readable signal: the server spots it in embeddings it already holds
Active label inference in vertical federated learning leaves a class-separability anomaly in the adversary's embeddings, letting the server trigger a defense that acts on the adversary alone and pushes label inference below what honest training already leaks on CIFAR-10, CIFAR-100 and CINIC-10, at under two points of task-accuracy cost.
Bangladesh University of Engineering and Technology
Soumya Swagata Biswas（First author）
Vertical Federated Learning Label Inference Attack Detection MaliciousSGD Embedding Asymmetry Defense
researchsquare.com Oct 14, 2026
Cross-day speech decoding without retraining the network: 65,792 input parameters lift all six T12 sessions
Across days, neural drift degrades a fixed intracortical speech decoder. One study confined adaptation to a 256×256 affine transform and bias at the neural input, updating 65,792 parameters, under 0.05% of the full model, while the recurrent encoder and phoneme classifier stayed frozen, using 40 or 80 labeled trials for supervised calibration. On the public T12 dataset, macro phoneme error rate fell 11.37% and 15.51% across six chronologically held-out sessions, with all six improving under all three seeds; on the independent T15 dataset the same frozen rule gave 2.72% and 10.05% reductions, with more session-to-session variability and neither budget below the nominal 0.05 reference threshold.
Across days, neural drift degrades a fixed intracortical speech decoder. One study confined adaptation to a 256×256 affine transform and bias at the neural input, updating 65,792 parameters, under 0.05% of the full model, while the recurrent encoder and phoneme classifier stayed frozen, using 40 or 80 labeled trials for supervised calibration. On the public T12 dataset, macro phoneme error rate fell 11.37% and 15.51% across six chronologically held-out sessions, with all six improving under all three seeds; on the independent T15 dataset the same frozen rule gave 2.72% and 10.05% reductions, with more session-to-session variability and neither budget below the nominal 0.05 reference threshold.
Rajshahi University of Engineering and Technology Bangladesh University of Engineering and Technology
Dibakar Roy（First author, Corresponding author）
Cross-session intracortical speech decoding Affine-only calibration Parameter-efficient adaptation Connectionist Temporal Classification
arXiv Sep 27, 2026
Reflex-Guard cuts prompt-safety filtering to 37.6 ms locally with dense semantic embeddings, reaching 95.9% harmful-prompt recall on 30,568 samples
The work introduces Reflex-Guard, a lightweight locally running guardrail for LLM prompt safety that combines jailbreak-aware preprocessing, compact sentence-transformer embeddings, and seven fast binary classifiers, achieving 95.9% recall on harmful prompts at 37.6 ms end-to-end latency on a balanced dataset of 30,568 samples drawn from five complementary sources, faster than Llama Guard 2 (255 ms) and SafeDecoding (723 ms), detecting 100% of GCG suffix attacks and Base64-encoded prompts at the default threshold, while DrAttack structured prompts required lowering the threshold to 0.03 for optimal detection.
The work introduces Reflex-Guard, a lightweight locally running guardrail for LLM prompt safety that combines jailbreak-aware preprocessing, compact sentence-transformer embeddings, and seven fast binary classifiers, achieving 95.9% recall on harmful prompts at 37.6 ms end-to-end latency on a balanced dataset of 30,568 samples drawn from five complementary sources, faster than Llama Guard 2 (255 ms) and SafeDecoding (723 ms), detecting 100% of GCG suffix attacks and Base64-encoded prompts at the default threshold, while DrAttack structured prompts required lowering the threshold to 0.03 for optimal detection.
New Mexico State University BRAC University Osaka Metropolitan University Bangladesh University of Engineering and Technology
Istiaque Ahmed（First author）
Reflex-Guard Sentence-transformer embeddings LLM prompt safety filtering
arXiv Sep 27, 2026
Reflex-Guard achieves 95.9% harmful-prompt recall at 37.6 ms locally, faster than Llama Guard 2 and SafeDecoding
The work introduces Reflex-Guard, a lightweight locally running guardrail for LLM prompt safety that combines jailbreak-aware preprocessing, compact sentence-transformer embeddings, and seven fast binary classifiers, achieving 95.9% recall on harmful prompts at 37.6 ms end-to-end latency on a strategically balanced dataset of 30,568 samples drawn from five complementary sources, faster than Llama Guard 2 (255 ms) and SafeDecoding (723 ms), detecting 100% of GCG suffix attacks and Base64-encoded prompts at the default threshold, while DrAttack structured prompts required lowering the threshold to 0.03 for optimal detection.
The work introduces Reflex-Guard, a lightweight locally running guardrail for LLM prompt safety that combines jailbreak-aware preprocessing, compact sentence-transformer embeddings, and seven fast binary classifiers, achieving 95.9% recall on harmful prompts at 37.6 ms end-to-end latency on a strategically balanced dataset of 30,568 samples drawn from five complementary sources, faster than Llama Guard 2 (255 ms) and SafeDecoding (723 ms), detecting 100% of GCG suffix attacks and Base64-encoded prompts at the default threshold, while DrAttack structured prompts required lowering the threshold to 0.03 for optimal detection.
New Mexico State University BRAC University Osaka Metropolitan University Bangladesh University of Engineering and Technology
Istiaque Ahmed（First author）
Reflex-Guard Sentence-transformer embeddings LLM prompt safety filtering

You’re all caught up

Flip the strand and DNA language models change their answer; MirrorDNA cuts both strands at mirrored points to give one prediction

SRAM sign-off stops simulating every corner: transferred design points plus conformal calibration still deliver 99.88% yield on a 1-Mb macro

Label theft in vertical federated learning now leaves a readable signal: the server spots it in embeddings it already holds

Cross-day speech decoding without retraining the network: 65,792 input parameters lift all six T12 sessions

Reflex-Guard cuts prompt-safety filtering to 37.6 ms locally with dense semantic embeddings, reaching 95.9% harmful-prompt recall on 30,568 samples

Reflex-Guard achieves 95.9% harmful-prompt recall at 37.6 ms locally, faster than Llama Guard 2 and SafeDecoding