Actor-Critic Methods: A2C and PPO

Actor-Critic Methods: A2C and PPO

Actor-Critic Architecture

Two networks work together:

• Actor: Policy π(a|s,θ_π) - decides which actions to take
• Critic: Value V(s,θ_v) - evaluates state quality

A2C (Advantage Actor-Critic)

class ActorCritic(nn.Module): def __init__(self, state_dim, action_dim, hidden_dim=128): super().__init__() # Shared backbone self.backbone = nn.Sequential( nn.Linear(state_dim, hidden_dim), nn.ReLU(), nn.Linear(hidden_dim, hidden_dim), nn.ReLU() ) # Actor head self.actor = nn.Sequential( nn.Linear(hidden_dim, hidden_dim), nn.ReLU(), nn.Linear(hidden_dim, action_dim), nn.Softmax(dim=-1) ) # Critic head self.critic = nn.Sequential( nn.Linear(hidden_dim, hidden_dim), nn.ReLU(), nn.Linear(hidden_dim, 1) ) def forward(self, state): backbone_output = self.backbone(state) action_probs = self.actor(backbone_output) state_value = self.critic(backbone_output) return action_probs, state_value

class A2CAgent: def __init__(self, state_dim, action_dim, lr_actor=1e-3, lr_critic=1e-3): self.network = ActorCritic(state_dim, action_dim).cuda() self.optimizer_actor = torch.optim.Adam(self.network.actor.parameters(), lr=lr_actor) self.optimizer_critic = torch.optim.Adam(self.network.critic.parameters(), lr=lr_critic) self.gamma = 0.99 def compute_advantage(self, states, rewards, next_states, dones): """Compute advantage: A(s,a) = r + γV(s') - V(s)""" states_tensor = torch.tensor(states, dtype=torch.float32).cuda() next_states_tensor = torch.tensor(next_states, dtype=torch.float32).cuda() rewards_tensor = torch.tensor(rewards, dtype=torch.float32).cuda() dones_tensor = torch.tensor(dones, dtype=torch.float32).cuda() with torch.no_grad(): _, values = self.network(states_tensor) _, next_values = self.network(next_states_tensor) # TD target targets = rewards_tensor + self.gamma * next_values.squeeze() * (1 - dones_tensor) advantages = targets - values.squeeze() return advantages, targets def train_on_batch(self, states, actions, rewards, next_states, dones): advantages, targets = self.compute_advantage(states, rewards, next_states, dones) states_tensor = torch.tensor(states, dtype=torch.float32).cuda() actions_tensor = torch.tensor(actions, dtype=torch.int64).cuda() # Forward pass action_probs, values = self.network(states_tensor) # Actor loss: -log π(a|s) * advantage log_probs = torch.log(action_probs.gather(1, actions_tensor.unsqueeze(1)).squeeze(1) + 1e-8) actor_loss = -(log_probs * advantages.detach()).mean() # Critic loss: MSE(V(s), target) critic_loss = ((values.squeeze() - targets.detach()) ** 2).mean() # Update actor self.optimizer_actor.zero_grad() actor_loss.backward() self.optimizer_actor.step() # Update critic self.optimizer_critic.zero_grad() critic_loss.backward() self.optimizer_critic.step() return actor_loss.item(), critic_loss.item() 

PPO (Proximal Policy Optimization)

Uses clipped surrogate objective to prevent large policy updates:

L^CLIP = 𝔼[min(r_t(θ) Â, clip(r_t(θ), 1-ε, 1+ε) Â)]

where r_t = π_new / π_old is the probability ratio.

class PPOAgent: def __init__(self, state_dim, action_dim): self.network = ActorCritic(state_dim, action_dim).cuda() self.optimizer = torch.optim.Adam(self.network.parameters(), lr=3e-4) self.gamma = 0.99 self.gae_lambda = 0.95 self.eps_clip = 0.2 def train_on_batch(self, states, actions, rewards, next_states, dones, old_log_probs, num_epochs=5): """ PPO training: sample multiple epochs on same batch with KL divergence constraint """ states_tensor = torch.tensor(states, dtype=torch.float32).cuda() actions_tensor = torch.tensor(actions, dtype=torch.int64).cuda() old_log_probs_tensor = torch.tensor(old_log_probs, dtype=torch.float32).cuda() # Compute advantages advantages, targets = self.compute_advantage(states, rewards, next_states, dones) advantages = (advantages - advantages.mean()) / (advantages.std() + 1e-8) for epoch in range(num_epochs): # Forward pass action_probs, values = self.network(states_tensor) # New log probabilities new_log_probs = torch.log( action_probs.gather(1, actions_tensor.unsqueeze(1)).squeeze(1) + 1e-8 ) # Probability ratio ratio = torch.exp(new_log_probs - old_log_probs_tensor) # Clipped surrogate loss surr1 = ratio * advantages surr2 = torch.clamp(ratio, 1 - self.eps_clip, 1 + self.eps_clip) * advantages actor_loss = -torch.min(surr1, surr2).mean() # Critic loss critic_loss = ((values.squeeze() - targets.detach()) ** 2).mean() # Total loss total_loss = actor_loss + 0.5 * critic_loss # Update self.optimizer.zero_grad() total_loss.backward() torch.nn.utils.clip_grad_norm_(self.network.parameters(), max_norm=0.5) self.optimizer.step() 

Why PPO Works

• Prevents catastrophic policy changes (clipping)
• More sample-efficient than A2C
• Simpler than TRPO (trust region policy optimization)
• State-of-the-art performance on continuous control tasks

Practice Problem

Implement both A2C and PPO. Train on continuous control (MuJoCo) environments. Compare sample efficiency, wall-clock training time, and final performance. Visualize learned policies. Implement generalized advantage estimation (GAE) for smoother advantage estimates.

Deep Dive: Actor-Critic Methods: A2C and PPO

At this level, we stop simplifying and start engaging with the real complexity of Actor-Critic Methods: A2C and PPO. In production systems at companies like Flipkart, Razorpay, or Swiggy — all Indian companies processing millions of transactions daily — the concepts in this chapter are not academic exercises. They are engineering decisions that affect system reliability, user experience, and ultimately, business success.

The Indian tech ecosystem is at an inflection point. With initiatives like Digital India and India Stack (Aadhaar, UPI, DigiLocker), the country has built technology infrastructure that is genuinely world-leading. Understanding the technical foundations behind these systems — which is what this chapter covers — positions you to contribute to the next generation of Indian technology innovation.

Whether you are preparing for JEE, GATE, campus placements, or building your own products, the depth of understanding we develop here will serve you well. Let us go beyond surface-level knowledge.

The Theory of Computation: What Can and Cannot Be Computed?

At the deepest level, computer science asks a philosophical question: what are the limits of computation? This leads us to some of the most beautiful ideas in all of mathematics:

  THE HIERARCHY OF COMPUTATIONAL PROBLEMS: ┌──────────────────────────────────────────────────┐ │ UNDECIDABLE — No algorithm can ever solve these  │ │ Example: Halting Problem │ │ "Will this program eventually stop or run │ │  forever?" — Alan Turing proved in 1936 that │ │  no general algorithm can determine this! │ ├──────────────────────────────────────────────────┤ │ NP-HARD — No known efficient algorithm │ │ Example: Travelling Salesman Problem │ │ "Visit all 28 state capitals with minimum │ │  travel distance" — checking all routes would │ │  take longer than the age of the universe │ ├──────────────────────────────────────────────────┤ │ NP — Verifiable in polynomial time │ │ P vs NP: Does P = NP? ($1 million prize!) │ ├──────────────────────────────────────────────────┤ │ P — Solvable efficiently (polynomial time) │ │ Examples: Sorting, searching, shortest path │ └──────────────────────────────────────────────────┘ If P = NP were proven, it would mean every problem whose solution can be VERIFIED quickly can also be SOLVED quickly. This would break all encryption, solve protein folding, and revolutionise science.

This is not just theoretical. The P vs NP question ($1 million Clay Millennium Prize) has profound implications: if P=NP, every encryption system in the world (including UPI, Aadhaar, banking) would be breakable. Indian mathematicians and computer scientists at ISI Kolkata, IMSc Chennai, and IIT Kanpur are actively researching computational complexity theory and related fields. Understanding these theoretical foundations is what separates a programmer from a computer scientist.

India's Scale Challenges: Engineering for 1.4 Billion

Building technology for India presents unique engineering challenges that make it one of the most interesting markets in the world. UPI handles 10 billion transactions per month — more than all credit card transactions in the US combined. Aadhaar authenticates 100 million identities daily. Jio's network serves 400 million subscribers across 22 telecom circles. Hotstar streamed IPL to 50 million concurrent viewers — a world record. Each of these systems must handle India's diversity: 22 official languages, 28 states with different regulations, massive urban-rural connectivity gaps, and price-sensitive users expecting everything to work on ₹7,000 smartphones over patchy 4G connections. This is why Indian engineers are globally respected — if you can build systems that work in India, they will work anywhere.

Transformer Architecture: The Engine Behind GPT and Modern AI

The Transformer architecture, introduced in the landmark 2017 paper "Attention Is All You Need," revolutionised NLP and eventually all of deep learning. Here is the core mechanism:

# Self-Attention Mechanism (simplified)
import numpy as np

def self_attention(Q, K, V, d_k): """ Q (Query): What am I looking for? K (Key): What do I contain? V (Value): What do I actually provide? d_k: Dimension of keys (for scaling) """ # Step 1: Compute attention scores scores = np.matmul(Q, K.T) / np.sqrt(d_k) # Step 2: Softmax to get probabilities attention_weights = softmax(scores) # Step 3: Weighted sum of values output = np.matmul(attention_weights, V) return output

# Multi-Head Attention: Run multiple attention heads in parallel
# Each head learns different relationships:
# Head 1: syntactic relationships (subject-verb agreement)
# Head 2: semantic relationships (word meanings)
# Head 3: positional relationships (word order)
# Head 4: coreference (pronoun → noun it refers to)

The key insight of self-attention is that every token can attend to every other token simultaneously (unlike RNNs which process sequentially). This parallelism enables efficient GPU training. The computational complexity is O(n²·d) where n is sequence length and d is dimension, which is why context windows are a major engineering challenge.

State-of-the-art developments include: sparse attention (reducing O(n²) to O(n·√n)), mixture of experts (MoE — activating only a subset of parameters per input), retrieval-augmented generation (RAG — grounding responses in external documents), and constitutional AI (alignment through principles rather than RLHF alone). Indian researchers at institutions like IIT Bombay, IISc Bangalore, and Microsoft Research India are actively contributing to these frontiers.

Research Frontiers and Open Problems in Actor-Critic Methods: A2C and PPO

Beyond production engineering, actor-critic methods: a2c and ppo connects to active research frontiers where fundamental questions remain open. These are problems where your generation of computer scientists will make breakthroughs.

Quantum computing threatens to upend many of our assumptions. Shor's algorithm can factor large numbers efficiently on a quantum computer, which would break RSA encryption — the foundation of internet security. Post-quantum cryptography is an active research area, with NIST standardising new algorithms (CRYSTALS-Kyber, CRYSTALS-Dilithium) that resist quantum attacks. Indian researchers at IISER, IISc, and TIFR are contributing to both quantum computing hardware and post-quantum cryptographic algorithms.

AI safety and alignment is another frontier with direct connections to actor-critic methods: a2c and ppo. As AI systems become more capable, ensuring they behave as intended becomes critical. This involves formal verification (mathematically proving system properties), interpretability (understanding WHY a model makes certain decisions), and robustness (ensuring models do not fail catastrophically on edge cases). The Alignment Research Center and organisations like Anthropic are working on these problems, and Indian researchers are increasingly contributing.

Edge computing and the Internet of Things present new challenges: billions of devices with limited compute and connectivity. India's smart city initiatives and agricultural IoT deployments (soil sensors, weather stations, drone imaging) require algorithms that work with intermittent connectivity, limited battery, and constrained memory. This is fundamentally different from cloud computing and requires rethinking many assumptions.

Finally, the ethical dimensions: facial recognition in public spaces (deployed in several Indian cities), algorithmic bias in loan approvals and hiring, deepfakes in political campaigns, and data sovereignty questions about where Indian citizens' data should be stored. These are not just technical problems — they require CS expertise combined with ethics, law, and social science. The best engineers of the future will be those who understand both the technical implementation AND the societal implications. Your study of actor-critic methods: a2c and ppo is one step on that path.

Key Vocabulary

Here are important terms from this chapter that you should know:

🏗️ Architecture Challenge

Design the backend for India's election results system. Requirements: 10 lakh (1 million) polling booths reporting simultaneously, results must be accurate (no double-counting), real-time aggregation at constituency and state levels, public dashboard handling 100 million concurrent users, and complete audit trail. Consider: How do you ensure exactly-once delivery of results? (idempotency keys) How do you aggregate in real-time? (stream processing with Apache Flink) How do you serve 100M users? (CDN + read replicas + edge computing) How do you prevent tampering? (digital signatures + blockchain audit log) This is the kind of system design problem that separates senior engineers from staff engineers.

The Frontier

You now have a deep understanding of actor-critic methods: a2c and ppo — deep enough to apply it in production systems, discuss tradeoffs in system design interviews, and build upon it for research or entrepreneurship. But technology never stands still. The concepts in this chapter will evolve: quantum computing may change our assumptions about complexity, new architectures may replace current paradigms, and AI may automate parts of what engineers do today.

What will NOT change is the ability to think clearly about complex systems, to reason about tradeoffs, to learn quickly and adapt. These meta-skills are what truly matter. India's position in global technology is only growing stronger — from the India Stack to ISRO to the startup ecosystem to open-source contributions. You are part of this story. What you build next is up to you.

Crafted for Class 10–12 • Reinforcement Learning • Aligned with NEP 2020 & CBSE Curriculum