REVIEW 7 cited by
Gradient Surgery for Multi-Task Learning
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
While deep learning and deep reinforcement learning (RL) systems have demonstrated impressive results in domains such as image classification, game playing, and robotic control, data efficiency remains a major challenge. Multi-task learning has emerged as a promising approach for sharing structure across multiple tasks to enable more efficient learning. However, the multi-task setting presents a number of optimization challenges, making it difficult to realize large efficiency gains compared to learning tasks independently. The reasons why multi-task learning is so challenging compared to single-task learning are not fully understood. In this work, we identify a set of three conditions of the multi-task optimization landscape that cause detrimental gradient interference, and develop a simple yet general approach for avoiding such interference between task gradients. We propose a form of gradient surgery that projects a task's gradient onto the normal plane of the gradient of any other task that has a conflicting gradient. On a series of challenging multi-task supervised and multi-task RL problems, this approach leads to substantial gains in efficiency and performance. Further, it is model-agnostic and can be combined with previously-proposed multi-task architectures for enhanced performance.
Forward citations
Cited by 7 Pith papers
-
Learning Adaptive Multi-Task Guidance, Navigation, and Control via Hypernetworks
A hypernetwork maps continuous physics-informed task embeddings to shared actor-critic weights, mastering four orbital GNC tasks and composing novel ones without retraining, with sim-to-real on a floating platform.
-
PACE: Polar Axis-Conditioned Estimation for PairUAV Relative Localization
A shared image-pair network beats a single-head baseline by giving heading and range their own decoder readouts—PACE's raw model scores 0.002460 on the PairUAV hidden test.
-
Lethe: Adapter-Augmented Dual-Stream Update for Persistent Knowledge Erasure in Federated Unlearning
A federated-unlearning method that keeps erased knowledge from coming back during continued training, reporting under 1% resurfacing in most tested settings.
-
Benchmarking Massively Parallelized Multi-Task Reinforcement Learning for Robotics Tasks
The paper introduces MTBench, a GPU-accelerated benchmark for massively parallel multi-task RL, and reports experiments suggesting on-policy methods outperform off-policy baselines while value learning limits MTRL per...
-
Training Dynamics Underlying Language Model Scaling Laws: Loss Deceleration and Zero-Sum Learning
Loss deceleration, a piecewise-linear break in log-log loss curves, is attributed to zero-sum learning where per-example gradients oppose one another, and scaling helps by mitigating it.
-
FA-LAM: Focus-Aware Large Avatar Model for One-Shot 4D Animatable Gaussian Head
A one-shot model for animatable 3D/4D Gaussian head reconstruction that adds attention regularization, decoupled reconstruction-animation training, and autoregressive visibility-gated fusion, reporting consistent metr...
-
Multi-Task Learning for Heterogeneous Prediction from Video Game State with Transfer Learning
On a large World of Tanks dataset, a shared multi-task model with equal weighting or PCGrad outperforms single-task models on average, and task/map pre-training helps most in low-data regimes.
Discussion (0). Sign in to comment.