More Like this

1. Booster: Tackling Harmful Fine-tuing for Large Language Models via Attenuating Harmful Perturbation,

   Huang, Tiansheng; Hu, Sihao; Ilhan, Fatih; Tekin, Selim F; Liu, Ling (April 2025, Proceedings of International Conference on Learning Representations (ICLR 2025))

   Harmful fine-tuning attack poses serious safety concerns for large language models’ fine-tuning-as-a-service. While existing defenses have been proposed to mitigate the issue, their performances are still far away from satisfactory, and the root cause of the problem has not been fully recovered. To this end, we in this paper show that harmful perturbation over the model weights could be a probable cause of alignment-broken. In order to attenuate the negative impact of harmful perturbation, we propose an alignment-stage solution, dubbed Booster. Technically, along with the original alignment loss, we append a loss regularizer in the alignment stage’s optimization. The regularizer ensures that the model’s harmful loss reduction after the simulated harmful perturbation is attenuated, thereby mitigating the subsequent fine-tuning risk. Empirical results show that Booster can effectively reduce the harmful score of the fine-tuned models while maintaining the performance of downstream tasks. Our code is available at https://github.com/git-disl/Booster. 

   more » « less
2. Biosignal Authentication Considered Harmful Today

   Krish, V; Paoletti, N; Kazemi, M; Smolka, S A; Rahmati, A (August 2024, USENIX)

   User authentication systems based on cardiovascular biosignals have gained prominence in recent years, as these signals are presumed to be difficult to forge. We challenge this assumption by showing that an observer who has access to one type of cardiac data --- such as a user's pulse waveform, readily obtainable from video and commercial smartwatches --- can design a spoofing attack strong enough to fool authentication systems based on other cardiovascular biosignals. We present BioForge, an approach that leverages a cycle-consistent generative adversarial network to synthesize realistic physiological signals for a given user without relying on simultaneously collected supervision data. We evaluate BioForge on multiple open-access datasets and an array of verification systems, many of which can be fooled over 50% of the time in 10 or fewer attempts. Notably, we are able to fool systems that rely not just on heart rate and peak locations but also on the morphology of the waveforms. We additionally showcase how BioForge can be used to spoof authentication systems from biosignal data extracted from video clips of a target user. Our work demonstrates that authentication systems should not rely on the secrecy of cardiovascular biosignals. 

   more » « less
3. Excessive SSD-Internal Parallelism Considered Harmful

   https://doi.org/10.1145/3599691.3603412  

   Zhang, Xiangqun; Pei, Shuyi; Choi, Jongmoo; Kim, Bryan S. (July 2023, ACM Workshop on Hot Topics in Storage and File Systems)

   Modern SSDs achieve high throughput by utilizing multiple independent channels and chips in parallel. However, we find that excessive parallelism inadvertently amplifies the garbage collection (GC) overhead due to the larger unit of space reclamation. Based on this observation, we design PLAN, a novel SSD parallelism management and data placement scheme that allocates different levels of parallelism to different workloads with different needs to minimize the GC overhead. We demonstrate the effectiveness of PLAN by evaluating it against other state-of-the-art designs across various real-world workloads. PLAN reduces write amplification with comparable or better performance to the other designs that are always at full parallelism. 

   more » « less
4. Wear leveling in SSDs considered harmful

   https://doi.org/10.1145/3538643.3539750  

   Jiao, Ziyang; Bhimani, Janki; Kim, Bryan S. (June 2022, Proceedings of the 14th ACM Workshop on Hot Topics in Storage and File Systems)

   We argue that wear leveling in SSDs does more harm than good under modern settings where the endurance limit is in the hundreds. To support this claim, we evaluate existing wear leveling techniques and show that they exhibit anomalous behaviors and produce a high write amplification. These findings are consistent with a recent large-scale field study on the operational characteristics of SSDs. We discuss the option of forgoing wear leveling and instead adopting capacity variance in SSDs, and show that the capacity variance extends the lifetime of the SSD by up to 2.94×. 

   more » « less
5. WTO must ban harmful fisheries subsidies

   https://doi.org/10.1126/science.abm1680  

   Sumaila, U. Rashid; Skerritt, Daniel J.; Schuhbauer, Anna; Villasante, Sebastian; Cisneros-Montemayor, Andrés M.; Sinan, Hussain; Burnside, Duncan; Abdallah, Patrízia Raggi; Abe, Keita; Addo, Kwasi A.; et al (October 2021, Science)