Model-based reinforcement learning approaches carry the promise of being data efficient. However, due to challenges in learning dynamics models that sufficiently match the real-world dynamics, they struggle to achieve the same asymptotic performance as model-free methods. We propose Model-Based Meta-Policy-Optimization…
On-device research index
arXiv research
A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
168,695 papers · 148 categories
Trend · papers per month
11 results for “MPO”
Researchers analyze how algorithmic and implementation choices affect RL performance.
problem Difficulty in separating algorithmic and implementation differences in RL performance.
method Unified derivations through a single control-as-inference objective, categorizing algorithms as EM or KL minimization.
result Implementation details are co-adapted with algorithmic choices, some transferable across algorithms.
New method compresses LSTM networks using MPS tensor trains.
problem Challenges in maintaining performance of compressed RNNs.
method Use of MPS tensor trains for LSTM network compression.
result MPS tensor trains outperform MPOs in storage and inference time.
We introduce a new algorithm for reinforcement learning called Maximum aposteriori Policy Optimisation (MPO) based on coordinate ascent on a relative entropy objective. We show that several existing methods can directly be related to our derivation. We develop two off-policy algorithms and demonstrate that they are com…
RL models outperform traditional methods in certain market conditions.
problem Traditional portfolio management methods rely on accurate forecasts and do not incorporate specific investor preferences.
method Deep reinforcement learning with specific investor preferences incorporated into reward functions, realistic transaction costs modelled.
result RL models can significantly outperform traditional methods in upward trending markets, but not in sideways trending markets.
Tensorized Rademacher projections outperform Gaussian projections in reducing tensor dimensions.
problem Reducing the dimension of high-dimensional tensors for machine learning.
method Tensorized Rademacher random projections using Tensor Train decomposition.
result Tensorized Rademacher projections can replace Gaussian projections in tensor compression.
Novel PO algorithms improve LLM alignment tasks.
problem Evaluating PO algorithms on LLM alignment is costly and noisy.
method Designed a diagnostic suite of MuJoCo tasks and datasets, used evolutionary strategies to discover specialized PO algorithms.
result Proposed MPO algorithms significantly outperform existing PO algorithms in LLM alignment.
We provide a framework for incorporating robustness -- to perturbations in the transition dynamics which we refer to as model misspecification -- into continuous control Reinforcement Learning (RL) algorithms. We specifically focus on incorporating robustness into a state-of-the-art continuous control RL algorithm call…
A new framework improves reinforcement learning algorithms with policy guarantees.
problem Designing efficient and stable reinforcement learning algorithms.
method A general framework (FMA-PG) based on functional mirror ascent that constructs surrogate functions enabling policy improvement guarantees.
result The proposed framework enables policy improvement guarantees that hold regardless of policy parameterization, and recovers important heuristics.
KL-constrained API shows optimization issues and improved with regularization.
problem Optimization issues in KL-constrained API algorithms.
method Comparison of KL divergence as a constraint vs. regularizer, empirical evaluation.
result KL-constrained API is not guaranteed to converge and incurs linear regret.
We present an off-policy actor-critic algorithm for Reinforcement Learning (RL) that combines ideas from gradient-free optimization via stochastic search with learned action-value function. The result is a simple procedure consisting of three steps: i) policy evaluation by estimating a parametric action-value function;…